Home Blog Page 7

The Ivalice Chronicles team had to remake the original Final Fantasy Tactics’ source code from scratch

0


The Square Enix team behind Final Fantasy Tactics — The Ivalice Chronicles didn’t just remaster the iconic strategy RPG, they had to go through the trouble of remaking the source code from scratch, according to Bloomberg. In an interview with Bloomberg‘s Jason Schreier, the game’s director Kazutoyo Maehiro explained the arduous process of designing The Ivalice Chronicles, which is set to release at the end of the month.

When getting to work on the remake, Maehiro and his team discovered they had to rebuild the source code from the ground up since it was lost thanks to the industry’s unstandardized practices in the ’90s, according to Bloomberg. When translating Final Fantasy Tactics from Japanese to English for the global release, the company would overwrite the original Japanese version’s code. For Maehiro, that meant the team had to undertake a ground-up overhaul and recreate the source code by playing the original game that released in 1997, consulting the game’s master disc and looking at the 2011 version called Final Fantasy Tactics: The War of the Lions, according to Bloomberg. Maehiro also said during a PAX West 2025 panel that the team gleaned a lot of lost info from fan-made databases.

Revealing more of the behind-the-scenes decisions for The Ivalice Chronicles, Maehiro told Bloomberg about the debate surrounding Count Cidolfus Orlandeau. Better known as Cid, and appropriately nicknamed Thunder God Cid, this overtuned character joins you later in the game, but many fans complained about him being overpowered. Instead of nerfing Cid, Maehiro told Bloomberg that keeping this character’s power level the same would better represent the storyline since “his role in the story is being that very powerful character who joins your party.” To quell any concerns of Cid being too broken, Maehiro told Bloomberg that the team decided to buff the other characters to even things out. Looking ahead, Maehiro also hinted at exploring sequels for the Final Fantasy Tactics franchise or even brand new games in the strategy RPG genre, given that The Ivalice Chronicles does well, according to Bloomberg.



Source link

Google details Gemini’s daily prompt and image edit usage limits

0


Google is finally going beyond “limited,” “expanded,” and “highest access” to detail Gemini’s usage limits, with actual numbers for free and paid customers.

Some of these usage limits have been known or at least suspected, but Google’s “Gemini Apps limits & upgrades for Google AI subscribers” support page is now explicit about everything.

Those on free accounts using Gemini 2.5 Pro have up to 5 prompts per day (with a 32,000 token context window). They can generate or edit up to 100 images daily, while Deep Research is limited to 5 reports per month. All users can generate 20 Audio Overviews per day.

Meanwhile, Google AI Pro subscribers ($19.99 per month) have 100 Gemini 2.5 Pro prompts, up to 1,000 images, and 3 Veo 3 Fast videos per day, while it’s 20 Deep Research reports per day.

Advertisement – scroll for more content

New vs. old (May)

Google AI Ultra customers ($249.99 per month) have up to 500 2.5 Pro prompts, 1,000 images (unchanged), and 5 Veo 3 videos per day, though it’s 200 Deep Research reports. They can send up to 10 Deep Think prompts/day with a 192,000 token context window.

Google says Canvas, Gems, and Storybook are “subject to the limits of the model that you select when you use the feature.”

More on Gemini:

FTC: We use income earning auto affiliate links. More.



Source link

Tornado warning expands to other parts of Western Mass. Saturday

0



Editor’s note: This story was updated at 2:36 p.m. Saturday.

A tornado warning has extended to other parts of Massachusetts after starting in New York Saturday.

The tornado warning was issued for southwestern Hampshire County and western Hampden County in Massachusetts Saturday. This includes Russell, Granville, Chester, Blandford, Tolland, Westfield, Huntington, Montgomery, Southwick, Southampton, Westhampton, Northampton and Easthampton. By 2:45 p.m. the storm is expected to be near Holyoke, West Springfield and Chesterfield. By 2:50 p.m. the storm is expected to be near Williamsburg.

The warning lasts until at least 3 p.m. Saturday.

There is also a tornado warning for northwestern Franklin County that lasts until 2:45 p.m.

At 2:31 p.m. the severe thunderstorm capable of producing a tornado was located over Shelburne, or near Greenfield, moving northeast at 40 mph.

A tornado warning was first issued in New York and the Berkshires Saturday.

The warning was originally issued for Northern Dutchess County in east central New York and Southwestern Columbia County in east central New York, indicated in red on the National Weather Service map. It has since expired.

It was later extended to Massachusetts in southwestern Berkshire County and Northwestern Litchfield County in northern Connecticut. This includes South Sandisfield, Otis Wood Lanes, Konkapot, New Boston, Algerie Four Corners, Bonny Rigg Corners, North Otis, Montville, FreeQuarter and Roosterville.

The warning lasts until 2:45 p.m.

“Take cover now,” the National Weather Service warned. “Move to a basement or an interior room on the lowestfloor of a sturdy building. Avoid windows. If you are outdoors, in amobile home, or in a vehicle, move to the closest substantial shelterand protect yourself from flying debris.”

Additionally, forecasters are warning of golf ball size hail.

At 1:55 p.m. the severe thunderstorm capable of producing a tornado was located near Copake, or 10 miles southwest of Great Barrington, moving northeast at 30 mph. At 2:15 p.m. the storm capable of producing a tornadowas located over Sandisfield, or 13 miles southeast of Great Barrington, moving east at 25 mph. Two minutes later, the severe thunderstorm capable of producing a tornadowas located over Sandisfield, or 12 miles southeast of Great Barrington, moving northeast at 45 mph.

Several counties across Massachusetts are under a severe thunderstorm watch.

The National Weather Service announced that the watch was put into effect at 1:10 p.m. and will stay until 8 p.m.

The following Massachusetts counties are on their radar:

  • Berkshire
  • Hampden
  • Norfolk
  • Essex
  • Hampshire
  • Suffolk
  • Franklin
  • Middlesex
  • Worcester

The storms are expected to be at its strongest in areas farther inland, particularly in Central and Western Massachusetts.

The greatest potential for thunderstorms could be anytime until 8 p.m., especially in the Worcester and Lowell regions.

For those closer to the coast, the threat is lower. The storms are expected to weaken as they approach the Boston and Providence corridor, where residents could see a few strong thunderstorms between 7 p.m. and 11 p.m.

Meanwhile, southeast Massachusetts may only experience a few rumbles of thunder.

Eastern Massachusetts, however, could be rain-free through the daylight hours. It’s very unlikely that a stray storm will pop up, the National Weather Service added.

“Whether you are heading to an outdoor event or hanging around the house, today is one of those days to remain weather aware,” the National Weather Service said. “Thankfully it looks like the storms will be quick to wind down after sunset.”

If you purchase a product or register for an account through a link on our site, we may receive compensation. By using this site, you consent to our User Agreement and agree that your clicks, interactions, and personal information may be collected, recorded, and/or stored by us and social media and other third-party partners in accordance with our Privacy Policy.



Source link

Davey Johnson, who won 2 World Series with Orioles, managed Mets to title, dies at 82 : NPR

0


Washington Nationals manager Davey Johnson laughs before a game against the Miami Marlins at Nationals Park in Washington, D.C. on Sept. 22, 2013.

Washington Nationals manager Davey Johnson laughs before a game against the Miami Marlins at Nationals Park in Washington, D.C. on Sept. 22, 2013. The All-Star second baseman, who won the World Series twice with the Baltimore Orioles as a player and managed the New York Mets to the title in 1986, has died.

Susan Walsh/AP


hide caption

toggle caption

Susan Walsh/AP

NEW YORK — Davey Johnson, an All-Star second baseman who won the World Series twice with the Baltimore Orioles as a player and managed the New York Mets to the title in 1986, has died. He was 82.

Longtime Mets public relations representative Jay Horwitz said Johnson’s wife, Susan, informed him of his death after a long illness. Johnson was at a hospital in Sarasota, Florida, when he died Friday, Horwitz said.

Johnson played 13 major league seasons with Baltimore, the Atlanta Braves, Philadelphia Phillies and Chicago Cubs from 1965-78 and won the Gold Glove three times while being voted an All-Star four times. He managed the Mets, Cincinnati Reds, Los Angeles Dodgers and the Washington Nationals during a span from 1984-2013.

Former Nationals general manager Mike Rizzo called it a tough day.

“Davey was a good man, close friend and a mentor,” Rizzo said in a text message. “A Hall of Fame caliber manager with a baseball mind ahead of his time.”

Darryl Strawberry, a member of the ’86 Mets, on social media called Johnson a remarkable leader who transformed the franchise into a winning organization.

“His ability to empower players to express themselves while maintaining a strong commitment to excellence was truly inspiring,” Strawberry posted on Instagram with a photo of him, Johnson and Dwight “Doc” Gooden. “Davey’s legacy will forever be etched in the hearts of fans and players alike. My heartfelt condolences go out to Susan Johnson and the entire Johnson family during this difficult time. He will be missed but remembered for his incredible impact on the game and the lives he touched.”

“He knew how to get the best out of everyone”

Ryan Zimmerman, who played for Johnson with Washington from 2011-13, said Johnson was an even better human than he was a baseball man.

“He knew how to get the best out of everyone — on and off the field,” Zimmerman said in a text message. “I learned so much from him, and my career would not have been the same without my years with him. He will be deeply missed by so many people.”

Johnson was AL manager of the year in 1997 when Baltimore won the division. He was NL manager of the year in 2012 when the Nationals made the playoffs for the first time since the move from Montreal.

“Davey was a world-class manager,” owner Mark Lerner said in a statement. “I’ll always cherish the memories we made together with the Nationals, and I know his legacy will live on in the heads and minds of our fans and those across baseball.”





Source link

The growing debate over age verification laws

0


Technologists and policymakers are reckoning with a generation-defining problem on the internet: while it can be a revolutionary force for unprecedented education and connection across the globe, it can also pose dangers to children when they have completely unfettered access.

There is no simple way, however, to monitor children’s internet access without surveilling adults, paving the way for disastrous online privacy violations.

While some advocates praise these laws as victories for children’s safety, many security experts warn that these laws are being proposed and passed with flawed implementation plans, which pose dangerous security risks for adult users as well.

Here’s a primer on where the debate over age and identity verification stands, and where these laws are being enacted.

What exactly is age verification?

When we talk about age verification laws, we aren’t talking about when you made a Neopets account as a kid and checked a box to affirm that you were at least 13 years old. In the United States, those types of age checks are a result of the Children’s Online Privacy Protection Act (COPPA), an internet safety law passed in 1998. But, as you already know, if you had a Neopets account when you were 10, COPPA-era age checks are very easy to navigate around. You simply click a box that says you’re 13.

In the context of the laws that have cropped up during the 2020s, age verification usually refers to a user uploading an official ID to a third-party verification system to prove who they are. Users might also upload biometric facial scans, like the ones that power Face ID on iPhones.

What is the point of age verification?

Of course, internet safety is not really about preventing children from playing games like Neopets. Parents and lawmakers are concerned about children accessing content that’s potentially dangerous for minors, like online pornography, information about illicit drug use, and social media sites where they may encounter strangers with bad intentions.

Techcrunch event

San Francisco
|
October 27-29, 2025

These concerns are not unfounded. Parents have turned to lawmakers to share horrific stories of how their children died after purchasing fentanyl-laced drugs on Facebook, or how they took their own lives after facing incessant bullying on Snapchat.

As technology becomes more sophisticated, the problem is getting worse: Meta’s AI chatbots have reportedly flirted with children, while Character.AI and OpenAI are facing lawsuits over the suicides of children that were allegedly encouraged by the companies’ chatbots.

We know the internet isn’t all bad, though. Without leaving your home or spending any money, you can learn to play guitar or write code. You can forge meaningful friendships with people from the other side of the world. You can access specialized telehealth care, even if you live somewhere where no doctor is trained in your diagnosis. You can find the answer to just about any question you want at any given moment (the capital of Madagascar is Antananarivo, by the way).

This is how global lawmakers have arrived at what they believe to be a sound compromise: they won’t nuke the whole internet, but they’ll just put certain content behind a gate that you can only unlock if you can prove you’re an adult. But in this case, you’re not just clicking a box to confirm your age – you’re uploading your government ID or scanning your biometric data to prove you can access certain content.

Is it safe to verify your identity by uploading a government ID or a biometric scan?

The safety of any digital security measure depends on its implementation.

Apple builds out products like Face ID so that these biometric scans of your face never leave your iPhone – they’re never shared over the cloud, which massively limits the potential for hackers to gain access.

But when any sort of connection to another network gets involved, that’s when identity verification can get fishy. We’ve already watched how these measures can play out poorly when the technology is anything but rock-solid.

“No method of age verification is both privacy-protective and entirely accurate,” the Electronic Frontier Foundation writes. “These methods don’t each fit somewhere on a spectrum of ‘more safe’ and ‘less safe,’ or ‘more accurate’ and ‘less accurate.’ Rather, they each fall on a spectrum of ‘dangerous in one way’ to ‘dangerous in a different way.’”

In recent memory, we have some strong examples of how badly things can go when a company slips up on its security.

On Tea, an app that women use to share information about men they meet on dating apps, users have to upload selfies and photos of their IDs to prove that they are who they say they are. But users on 4chan, a misogynistic web forum, found that Tea left users’ data exposed, meaning that bad actors could access tens of thousands of users’ government IDs, selfies, and even direct messages on the platform, where women shared sensitive information about their dating experiences. What was once purported to be an app for women’s safety ended up exposing its users to vicious harassment, giving bad actors access to personal information like their home addresses.

These hacks were possible despite Tea’s promise that these images were not stored anywhere and were deleted immediately (evidently, those claims were false).

This kind of thing happens all the time – just look at TechCrunch’s security coverage. But it’s not just happening to new apps like Tea. World governments and trillion-dollar tech giants are certainly not exempt from data breaches.

Does it really matter if I lose my anonymity on the internet? I’m not doing anything shady.

These laws have inspired much backlash, but it’s not just because people are shy about linking their porn viewership to their government IDs.

In places where people can be prosecuted for political speech, anonymity is vital to allow people to meaningfully discuss current events and critique those in power without fear of retribution. Corporate whistleblowers could be unable to report a company’s wrongdoing if all of their online activity is linked to their identity, and victims of domestic abuse will find it even more difficult to flee dangerous situations.

In the U.S., the idea of being prosecuted for one’s political beliefs is becoming less theoretical. President Trump has threatened to send his political opponents to prison, and the government has revoked visas from international students who have criticized the Israeli government or participated in protests against the country’s military actions.

What age verification laws have gone into effect in the U.S.?

In the United States, twenty-three states have enacted age verification laws as of August 2025, while two more states have laws slated to take effect in late September 2025.

These laws mostly impact websites that host certain percentages of “sexual material harmful to minors,” which varies from state to state.

In practice, this means that pornographic websites must verify a user’s identity before they can access the website. But some sites, like Pornhub, have opted to simply block traffic from certain states.

“Since age verification software requires users to hand over extremely sensitive information, it opens the door to the risk of data breaches,” Pornhub wrote on its blog. “Whether or not your intentions are good, governments have historically struggled to secure this data.”

What counts as “sexual material harmful to minors”?

The definition of this term varies depending on who is enforcing the law.

At a time when LGBTQ rights are under attack in the U.S., activists have warned that laws like this could be used to classify non-pornographic information about the LGBTQ community, as well as basic sex education, as “sexual material harmful to minors.” These concerns appear well-founded, given that President Trump’s administration has removed references to civil rights movements and LGBTQ history from some government websites.

Texas’s age verification law – which was upheld in a Supreme Court ruling in June – was passed around the same time the state imposed other legal restrictions on the LGBTQ community, including limits on public drag shows and bans on gender-affirming care for minors. The drag show law was later deemed unconstitutional for violating the First Amendment.

What’s going on with age verification in the U.K.?

The United Kingdom enacted the Online Safety Act in July 2025, requiring many online platforms to verify a user’s identity before allowing them access. If a user is identified as a minor, they won’t be allowed on certain websites. The Act applies to search engines, social media platforms, video-sharing platforms, instant messaging services, cloud storage sites – pretty much anywhere that you may encounter media or talk to someone.

In practice, this means that websites like YouTube, Spotify, Google, X, and Reddit are requiring UK users to verify their identity before accessing certain content. These requirements don’t just apply to pornographic or violent content – people in the UK have been barred from viewing vital education and news sources, making it difficult to access information without exposing themselves to potential privacy concerns.

The UK does not use one specific way of verifying one’s identity – individual websites can decide what mechanism to use, and Ofcom, the UK’s communications regulator, is supposed to oversee this implementation. But as we explained with the Tea example, we can’t trust that any given authentication tool will be safe.

Now, users who are subject to identity verification must decide if they want to freely access information or if they want to expose themselves to privacy risks.

Does the U.K. age verification law affect me if I live elsewhere?

Even if you don’t live in the U.K., you may be impacted by tech platforms that are pre-complying with these regulations.

In the U.S., YouTube has already begun to roll out technology that is supposed to estimate users’ ages based on their activity, regardless of what age they listed when registering their account.

Can’t you just use a VPN to get around these barriers?

Yes, and the App Store charts in the U.K. prove it – after the Online Safety Act took effect, half of the top ten free apps on iOS were VPNs (Virtual Private Networks). We also saw VPN downloads spike after Pornhub access was blocked in many U.S. states.

When Pornhub was suspended in France, ProtonVPN said that registrations had spiked by 1000% within half an hour – the company said this was a bigger spike than when TikTok temporarily blocked American users.

You may have used a VPN before if you logged into your office desktop computer remotely, or if you spoofed your location so that you could watch British sitcoms for free from the U.S.

This introduces another issue: free VPNs don’t always have great privacy practices, even if they are advertised as such.

If you want to learn more about VPNs, TechCrunch has guides on what you need to know about VPNs and how you can decide if you need to use one.



Source link

Unity developers can now tap into system screen reader tools on macOS and Windows

0


Unity is to support native screen readers in both macOS and Windows. The feature is available now in the Unity 6000.3.0a5 alpha, and should make the process of making games accessible for blind players cheaper for developers, .

Screen readers narrate on-screen menus so blind and low-vision players can navigate a game or a piece of software without additional assistance. Typically, screen reading software is custom-built for each game, which can make them resource-intensive for developers to implement. “Building something like that from scratch has to be decided upon early in development so you have the time/resources allocated to make it properly,” Steve Saylor, an accessibility consultant and creator, . “Having it in-engine can mean the heavy lifting is done for you, and the cost of time/resources now is significantly lower.”

Unity previously offered APIs for both Android and iOS’ built-in screen readers in its Unity 6.0 release, but hadn’t yet added support for Windows Narrator or macOS VoiceOver. With this new alpha and its eventual release as Unity 6.3, developers creating games with Unity will have access to a native screen reader in all of the engine’s major platforms. Considering how popular Unity is as a game engine, that could vastly improve the accessibility of future games.



Source link

Samsung One UI 8 rollout schedule leaks for Galaxy phones

0


While One UI 8 is already available on some Samsung Galaxy devices, a new schedule finally offers some insight as to when the company will be rolling out the Android 16 update to more of its catalog.

If you have a modern Galaxy phone, the company’s take on Android 16 will arrive at some point. But Samsung’s update schedule can certainly vary – and by quite a lot – from device to device.

A schedule from Samsung has surfaced an early schedule for the rollout of One UI 8 and Android 16 for a large list of Galaxy devices, giving us an idea of what to expect in the coming months. These dates will likely vary between regions, and are claimed to mostly focus on Asia.

That said, here’s the list as it stands today (via SammyGuru):

Advertisement – scroll for more content

  • Galaxy S25 Ultra – September 18
  • Galaxy S25+ – September 18
  • Galaxy S25 – September 18
  • Galaxy S25 Edge – September 25
  • Galaxy S24 Ultra – September 25
  • Galaxy S24+ – September 25
  • Galaxy S24 FE – September 25
  • Galaxy S24 – September 25
  • Galaxy A56 5G – September 25
  • Galaxy A36 5G – September 25
  • Galaxy Tab S10 Ultra 5G – October 1
  • Galaxy Tab S10 Ultra – October 1
  • Galaxy Tab S10+ 5G – October 1
  • Galaxy Tab S10+ – October 1
  • Galaxy Z Fold 6 – October 2
  • Galaxy Z Flip 6 – October 2
  • Galaxy S23 Ultra – October 2
  • Galaxy S23+ – October 2
  • Galaxy S23 FE – October 2
  • Galaxy S23 – October 2
  • Galaxy S21 FE 5G – October 2
  • Galaxy A26 5G – October 2
  • Galaxy A17 5G – October 2
  • Galaxy A16 5G – October 2
  • Galaxy S22 Ultra – October 6
  • Galaxy S22+ – October 6
  • Galaxy S22 – October 6
  • Galaxy Z Fold 4 – October 6
  • Galaxy Z Flip 4 – October 6
  • Galaxy A55 5G – October 6
  • Galaxy Tab S10 FE+ 5G – October 9
  • Galaxy Tab S10 FE+ – October 9
  • Galaxy Tab S10 FE 5G – October 9
  • Galaxy Tab S10 FE – October 9
  • Galaxy Z Fold 5 – October 13
  • Galaxy Z Flip 5 – October 13
  • Galaxy A54 5G – October 13
  • Galaxy A52s 5G – October 13
  • Galaxy A25 5G – October 16
  • Galaxy A23 5G – October 16
  • Galaxy Tab Active 5 5G – October 16
  • Galaxy Tab Active 5 – October 16
  • Galaxy M34 5G – October 20
  • Galaxy Tab S9 Ultra – October 23
  • Galaxy Tab S9+ – October 23
  • Galaxy Tab S9 – October 23
  • Galaxy Tab S8 Ultra – October 23
  • Galaxy Tab S8+ – October 23
  • Galaxy Tab S8 – October 23
  • Galaxy Tab S6 Lite – October 23
  • Galaxy XCover 7 – October 23
  • Galaxy XCover 6 Pro – October 23
  • Galaxy M33 5G – October 27
  • Galaxy M15 5G – October 27
  • Galaxy A53 5G – October 30
  • Galaxy A35 5G – October 30
  • Galaxy A34 5G – October 30
  • Galaxy A33 5G – October 30
  • Galaxy Tab A9 – November 5
  • Galaxy Tab Active 5 Pro 5G – November 7
  • Galaxy XCover 7 Pro – November 10

There are some notable oddities here, such as the Galaxy Z Fold 6 and Flip 6 being updated after the Galaxy S23 series, and the Galaxy S21 FE being updated before the Galaxy S22 series.

Beyond that, though, Samsung is also planning some One UI 8 Watch updates for Galaxy Watch models, which break down as follows:

  • Galaxy Watch 7 – October 1
  • Galaxy Watch 6 Classic – October 1
  • Galaxy Watch 6 – October 1
  • Galaxy Watch FE – October 1
  • Galaxy Watch 5 Pro – November 3
  • Galaxy Watch 5 – November 3
  • Galaxy Watch 4 Classic – November 3
  • Galaxy Watch 4 – November 3

More on Samsung:

Follow Ben: Twitter/XThreads, Bluesky, and Instagram

FTC: We use income earning auto affiliate links. More.





Source link

3 Monkeys continues to serve up flavor and fun in Harwich Port

0



A longtime staple in the Harwich Port dining scene, 3 Monkeys Street Bar + Kitchen + Sushi continues to delight Cape Cod locals and tourists with its bold flavors, playful atmosphere and open-air dining experience.

Located at 554 Route 28, the restaurant has carved out a unique niche, offering a blend of Japanese street food, cocktails and a laid-back coastal charm.

3 Monkeys isn’t your average Cape seafood shack. It brings a fresh perspective to Cape Cod cuisine, serving up sushi, ramen, bao and noodle dishes alongside a dynamic bar menu.

The colorful, open-air space gives diners the opportunity to enjoy food and drink in a setting that feels both trendy and relaxed; ideal for everything from casual lunches to lively nights out.

One of 3 Monkeys’ biggest draws is its inclusive approach to dining. The kitchen readily accommodates dietary needs with vegan, vegetarian and gluten-free options available upon request.

Beyond the food, 3 Monkeys is also known for its welcoming environment and community spirit.

The restaurant offers private event bookings through its website and frequently serves as a hotspot for social gatherings during Cape Cod’s bustling summer season. The festive, come-as-you-are vibe makes it a go-to for both locals and out-of-towners.

If you purchase a product or register for an account through a link on our site, we may receive compensation. By using this site, you consent to our User Agreement and agree that your clicks, interactions, and personal information may be collected, recorded, and/or stored by us and social media and other third-party partners in accordance with our Privacy Policy.



Source link

OpenAI reorganizes research team behind ChatGPT’s personality

0


OpenAI is reorganizing its Model Behavior team, a small but influential group of researchers who shape how the company’s AI models interact with people, TechCrunch has learned.

In an August memo to staff seen by TechCrunch, OpenAI’s chief research officer Mark Chen said the Model Behavior team — which consists of roughly 14 researchers — would be joining the Post Training team, a larger research group responsible for improving the company’s AI models after their initial pre-training.

As part of the changes, the Model Behavior team will now report to OpenAI’s Post Training lead Max Schwarzer. An OpenAI spokesperson confirmed these changes to TechCrunch.

The Model Behavior team’s founding leader, Joanne Jang, is also moving on to start a new project at the company. In an interview with TechCrunch, Jang says she’s building out a new research team called OAI Labs, which will be responsible for “inventing and prototyping new interfaces for how people collaborate with AI.”

The Model Behavior team has become one of OpenAI’s key research groups, responsible for shaping the personality of the company’s AI models and for reducing sycophancy — which occurs when AI models simply agree with and reinforce user beliefs, even unhealthy ones, rather than offering balanced responses. The team has also worked on navigating political bias in model responses and helped OpenAI define its stance on AI consciousness.

In the memo to staff, Chen said that now is the time to bring the work of OpenAI’s Model Behavior team closer to core model development. By doing so, the company is signaling that the “personality” of its AI is now considered a critical factor in how the technology evolves.

In recent months, OpenAI has faced increased scrutiny over the behavior of its AI models. Users strongly objected to personality changes made to GPT-5, which the company said exhibited lower rates of sycophancy but seemed colder to some users. This led OpenAI to restore access to some of its legacy models, such as GPT-4o, and to release an update to make the newer GPT-5 responses feel “warmer and friendlier” without increasing sycophancy.

Techcrunch event

San Francisco
|
October 27-29, 2025

OpenAI and all AI model developers have to walk a fine line to make their AI chatbots friendly to talk to but not sycophantic. In August, the parents of a 16-year-old boy sued OpenAI over ChatGPT’s alleged role in their son’s suicide. The boy, Adam Raine, confided some of his suicidal thoughts and plans to ChatGPT (specifically a version powered by GPT-4o), according to court documents, in the months leading up to his death. The lawsuit alleges that GPT-4o failed to push back on his suicidal ideations.

The Model Behavior team has worked on every OpenAI model since GPT-4, including GPT-4o, GPT-4.5, and GPT-5. Before starting the unit, Jang previously worked on projects such as Dall-E 2, OpenAI’s early image-generation tool.

Jang announced in a post on X last week that she’s leaving the team to “begin something new at OpenAI.” The former head of Model Behavior has been with OpenAI for nearly four years.

Jang told TechCrunch she will serve as the general manager of OAI Labs, which will report to Chen for now. However, it’s early days, and it’s not clear yet what those novel interfaces will be, she said.

“I’m really excited to explore patterns that move us beyond the chat paradigm, which is currently associated more with companionship, or even agents, where there’s an emphasis on autonomy,” said Jang. “I’ve been thinking of [AI systems] as instruments for thinking, making, playing, doing, learning, and connecting.”

When asked whether OAI Labs will collaborate on these novel interfaces with former Apple design chief Jony Ive — who’s now working with OpenAI on a family of AI hardware devices — Jang said she’s open to lots of ideas. However, she said she’ll likely start with research areas she’s more familiar with.

This story was updated to include a link to Jang’s post announcing her new position, which was released after this story published. We also clarify the models that OpenAI’s Model Behavior team worked on.





Source link

Zuckerberg caught on hot mic telling Trump ‘I wasn’t sure’ how much to promise to spend on AI in the US

0


Mark Zuckerberg has certainly come a long way in his relationship with President Donald Trump. Almost exactly a year after the president threatened the Meta CEO with imprisonment, the two sat side-by-side at a White House dinner, alongside numerous other tech CEOs.

The nearly three dozen CEOs and execs in attendance took turns praising and thanking Trump. But Zuckerberg’s comments were especially notable. In one moment that was widely shared on social media, Trump turns to Zuckerberg and asks “how much are you spending, would say, over the next few years?” Zuckerberg responded that it was “probably going to be something like, I don’t know, at least $600 billion through [20]28 in the US.” Trump seemed to approve. “That’s a lot, thank you Mark, it’s great to have you.”  

But it was a hot mic moment captured later between the two that was especially telling. Zuckerberg, turning to Trump, apologizes and says “sorry, I wasn’t ready …I wasn’t sure what number you wanted to go with.” 

You can watch the whole moment play out in the clip below:

While Zuckerberg has spent the last year trying to curry favor with Trump, their interactions show just how much those efforts have been paying off. A year ago, the then-former president was threatening the Facebook founder with jail time. Now, after donating $1 million to his inauguration, changing Meta’s policies and renouncing DEI, adding a pro-Trump booster to his board, paying $25 million to settle a four-year-old lawsuit  and several private meetings, the two seem to have patched things up. Not only is Zuckerberg promising to spend massive amounts on money in the US on AI infrastructure, he’s seemingly confirming that Trump approves of the specific number.



Source link