On Monday, OpenAI and its CEO Sam Altman were sued by the Florida attorney general in a first-of-its-kind state lawsuit over ChatGPT’s alleged ties to a number of violent incidents.

The lawsuit accuses OpenAI of diverting attention from security as it seeks to prioritize winning the “AI arms race and amassing vast fortunes.”

“Today we announced the nation’s first state-led lawsuit against OpenAI and its CEO Sam Altman,” said Florida Attorney General James Uthmeier. “OpenAI and Altman ignored internal and external security warnings, exposed children to enormous risk, and allowed an unsafe product to reach millions of Floridians.”

“Because of defendants’ false statements about ChatGPT and their careless introduction of ChatGPT to Florida and the world, mass shooters were aided and abetted in deadly rampages, vulnerable people were encouraged to commit suicide, professionals suffered public humiliation, users lost their critical thinking skills, and minors became addicted to a tool masquerading as human compassion to collect data without parental supervision,” he said. 83-page lawsuit claims.

In April, the Florida Attorney General’s office opened an investigation into the company. The purpose of this investigation was to determine what role ChatGPT may have played in the mass shooting that occurred last year at Florida State University. It is alleged that the shooter consulted a chatbot before the attack. OpenAI was there too defendant in a civil suit by the family of one of the victims of this shooting.

OpenAI has previously denied responsibility for the Florida shooting. “Last year’s mass shooting at Florida State University was a tragedy, but ChatGPT is not responsible for this terrible crime” – OpenAI spokesperson he previously told NBC News. TechCrunch has reached out to OpenAI for comment.

OpenAI just settled another legal case involving former co-founder Elon Musk, who sued the company in 2024, accusing it of betraying its original mission to assist humanity by turning the organization into a for-profit business. The case ended after a jury quickly decided that Musk waited too long to file the lawsuit and that the statute of limitations had run out.

This is just the latest legal case that has attempted to link ChatGPT to violent deaths. Last year, OpenAI was sued by the parents of Adam Raine, a California teenager who took his own life after talking to a chatbot about suicide. In this case, ChatGPT allegedly offered “technical specifications” for various suicide methods, although it also referred him to mental health resources. Other lawsuits — including those alleging chatbot liability for suicides, stalking, and murder – are in progress.

OpenAI claims his up-to-date model of reasoning provided an original mathematical proof refuting the famed unsolved hypotheses in geometry first posed by Paul Erdős in 1946.

If this sounds familiar to you, it’s because this isn’t the first time OpenAI has made such a bold claim. Seven months ago, former vice president of the artificial intelligence giant, Kevin Weil, posted on X: “GPT-5 found solutions to 10 (!) previously unsolved Erdő problems and made progress on 11 others.”

It turns out that GPT-5 didn’t actually solve these problems; he simply found solutions that already existed in the literature.

Taunts from rivals like Yann LeCun and Google DeepMind CEO Demis Hassabis followed, and Weil immediately deleted his premature post. At least today it seems that OpenAI hasn’t made the same mistake twice. Along with the announcement, the company published comrade’s comments in support of a rebuttal from mathematicians such as Noga Alon, Melanie Wood, and Thomas Bloom who hold on the Erdos Problems pageand formerly called the Weil brisk “dramatic misrepresentation”.

“For almost 80 years, mathematicians believed that the best possible solutions looked something like square grids.” OpenAI published on X. “The OpenAI model has debunked this belief by discovering an entirely new family of designs that perform better.”

The company stated that this is “the first time that artificial intelligence has independently solved a significant open problem of key importance in the field of mathematics.” According to OpenAI, the proof comes from a up-to-date general-purpose reasoning model, not from a system designed specifically to solve math problems or even this specific problem.

OpenAI says this is significant because it means AI systems are now better able to connect long, challenging chains of reasoning and connect ideas from different fields in ways that researchers may not have explored before. This has implications for biology, physics, engineering and medicine.

“Artificial intelligence is helping us more fully understand the mathematics department we have spent centuries building,” Bloom said in a statement. “What other unseen wonders are waiting in the wings?”

OpenAI co-founder and CEO Greg Brockman officially takes the helm of the company’s product strategy, according to Wired.

This appears to reinforce a change already in place, as Brockman is temporarily overseeing OpenAI’s products while the company’s AGI implementation CEO, Fidji Simo, is on diseased leave. Wired also reports that in a staff memo, Brockman described plans to combine ChatGPT and its Codex software product into a single, unified solution.

“We are consolidating our product efforts to ensure they are delivered with maximum focus on the agent future to drive success with both consumers and enterprises,” Brockman reportedly said.

OpenAI told TechCrunch that while Simo remains on medical leave, it has been working with Brockman on these changes. It was also noted that OpenAI has I’ve already talked about plans to combine ChatGPT, Codex and its APIs into one platform with one core product team.

Delayed last year, CEO Sam Altman declared a “code red” and stated that the company needed to refocus on the core ChatGPT experience. OpenAI has since paused “side quests” including the Sora video generator and OpenAI for Science.

On Friday, OpenAI launched a fresh set of personal finance tools in preview for ChatGPT Pro subscribers in the U.S., allowing them to connect accounts and ask questions ranging from spending analysis to future financial planning.

OpenAI has partnered with financial linking service Plaid to manage account linking. Users can connect with over 12,000 financial institutions, including Schwab, Fidelity, Chase, Robinhood, American Express and Capital One. Once users connect these accounts, they’ll see a dashboard showing their wallet performance, spending, subscriptions, and upcoming payments.

The fresh product comes just a month after OpenAI in April acquired the team behind personal finance startup Hiro, which is backed by companies including Ribbit, General Catalyst and Restive. OpenAI said Hiro’s team’s expertise in finance was useful in launching the product, but did not specify whether the entire feature was built by them.

OpenAI users can access the tool by selecting “Get Started” from the “Finance” option in the sidebar or by typing “@Finances, connect my accounts” in a ChatGPT conversation. Once users do this, the chatbot will instruct them on how to connect their accounts through Plaid. The company said it plans to support Intuit soon, which will enable analysis of, among others, the tax impact of selling shares or your chances of getting a credit card approved.

According to OpenAI, over 200 million users already ask ChatGPT financial questions every month. The company also noted that the fresh GPT-5.5 model better handles context, which is crucial when answering finance-related questions. The company said it worked with financial experts to create a benchmark for the model to improve personal finance issues.

With the fresh integration of financial tools, users can get detailed answers to questions such as: “I feel like I’ve been spending more lately. Has anything changed?” or “Help me develop a plan that will help me be ready to buy a home in my area in the next 5 years.”

Users can, if they wish, go to Settings > Apps > Finance and remove connections to specific accounts. Once the service is disconnected, your synchronized data will be deleted from ChatGPT within 30 days. Moreover, users can also view and delete financial memories on the Finance page.

Generalized chatbots are designed to answer everything, leading people to ask questions about data-sensitive topics such as health, finances and personal lives. AI companies realize this and produce specialized products for these sectors. Both OpenAI and Anthropic have launched health-related tools. Earlier this month, Perplexity launched its own version financial research product based on a computer agent.

OpenAI says its personal finance tools will be available on ChatGPT on the web and on iOS for Pro users. It noted that based on the feedback from these users, it wants to improve the product before making it available to Plus users.

Codex goes mobile. The coding tool — which OpenAI launched about a year ago — has now been integrated with the ChatGPT app, allowing users to remotely monitor and manage their development workflow.

The fresh feature allows users to see their Codex environments live on any device it is running on. The company announced the changes on Thursday; The update, which is currently in preview, is now available for all iOS and Android plans.

“It’s more than just the ability to remotely control a single task or send new tasks to a computer,” OpenAI said in a statement. “On your phone, you can work on all threads, view results, approve commands, change models, or start something new.”

Last month, OpenAI enabled Codex to run in the background in desktop environments, allowing it to perform various tasks on its own. Earlier this month, the company also introduced Chrome extension that allows the agent to work in energetic browser sessions.

Back in February, Anthropic released a similar feature – Remote control — which allows users to remotely monitor Claude Code’s work from afar.

The plethora of fresh features from both OpenAI and Anthropic demonstrates the tense competition between them over which agent coding tool will become the most widely adopted. Over the past year, Anthropic’s Claude Code tool has grown in popularity among both businesses and tech professionals, although both tools are still widely used.

OpenAI on Thursday announced a novel feature called Trusted Contact, designed to alert a trusted third party if self-harm is mentioned in a conversation. This feature allows an adult ChatGPT user to designate another person as a trusted contact on their account, such as a friend or family member. In cases where a conversation may escalate into self-harm, OpenAI will now prompt the user to contact that person. It also sends an automatic alert to the contact, prompting them to contact you.

OpenAI faced a wave of lawsuits from the families of people who committed suicide after talking to a chatbot. In many cases, families say ChatGPT encouraged a loved one commit suicide – or even helped them plan it.

Currently, OpenAI uses a combination of automation and manual control to handle potentially harmful incidents. Certain conversational triggers alert the company’s system to suicidal thoughts, which then relays the information to the human safety team. The company says that every time it receives this type of report, the incident is investigated by a human. “We aim to review these security alerts in less than an hour,” the company says.

If OpenAI’s internal team deems a situation to be a significant security risk, ChatGPT sends a notification to a trusted contact – via email, text, or in-app notification. The notification should be compact and encourage the contact person to contact the person. The company says it does not provide detailed information on the topic discussed to protect user privacy.

The Trusted Contact feature is compliant with company security introduced in September last year this gave parents the ability to have some oversight of their teens’ accounts, including collections security notifications designed to alert parents if the OpenAI system deems their child is facing a “serious safety risk”. For some time now, ChatGPT has also included automatic notifications to seek professional healthcare if the conversation turns towards the topic of self-harm.

Most importantly, Trust Contact is optional and even if protection is activated on a specific account, each user can have multiple ChatGPT accounts. OpenAI Parental Controls are also optional and come with a similar limitation.

“Trusted Contact is part of OpenAI’s broader efforts to build artificial intelligence systems help people in difficult times” – wrote the company in announcement post. “We will continue to work with clinicians, researchers and policymakers to improve the way AI systems respond when people may be experiencing anxiety.”

Etsy announced The native app within ChatGPT launched on Tuesday, opening up a modern way for buyers to browse its catalog of more than 100 million listings.

The modern experience aims to go beyond the limitations of conventional keyword-based queries. Instead of typing in “wooden coffee table” and then scrolling through and adjusting filters, users can now express what they’re looking for in natural language. For example: “Help me find a Mother’s Day gift under $100 for my mom who loves gardening.”

This feature, currently in beta, allows users to tag @Etsy directly in their confirmation. From there, the Etsy app in ChatGPT displays relevant product listings that users can browse, compare, and click to go to Etsy for more information or to purchase.

This isn’t Etsy’s first experiment with ChatGPT. In September, Etsy became an early ChatGPT partner Instant payment integration that allows users to purchase products directly within the chat interface. However, the initiative ended in Marchwhich suggests it didn’t work as OpenAI expected. It was reported that Etsy wasn’t seeing a lot of sales volume as a result of the integration, which led Etsy to start building a native app within ChatGPT.

In parallel with this launch, Etsy also revealed that it is testing a beta version of conversational search on its platform, aimed at helping users find gifts. The Gift Assistant acts as a personal shopper, offering a guided, conversational way to discover ideas, narrow down preferences and find the right products.

This builds on Etsy’s broader AI efforts, which it includes AI-powered discovery with selected collections and a package seller’s toolsincluding a tool that helps generate product titles and descriptions, as well as a writing assistant that helps prepare messages to buyers. In 2024, Etsy introduced a modern “Designed” label for AI-powered content as part of an effort to augment transparency as AI-generated art becomes more popular on the platform.

News of ChatGPT integration comes a week after Etsy reported first-quarter 2026 earnings exceeded revenue expectations at $631 million, and gross merchandise sales in the marketplace increased 6% year-over-year. It is worth noting that the number of energetic buyers increased for the first time in two years to 86.6 million. Etsy also boasted 5.6 million energetic sellers on the platform.

In February, the company announced it was selling Depop to eBay for $1.2 billion in cash in a move aimed at doubling its share of the mainstream market.

Etsy joins a growing list of companies building native apps on ChatGPT, including Angi, SeatGeek, Tubi and Wix. From October, developers can create applications within the chatbot.

On Tuesday, OpenAI released a novel core model called GPT-5.5 Instant, which will replace GPT-5.3 Instant as the default ChatGPT model. The company said the model reduces hallucinations in sensitive areas such as law, medicine and finance, while maintaining low latency compared to its predecessor.

Last month, OpenAI released its latest GPT-5.5 model, with the company claiming it has improved in areas such as coding and knowledge work.

The novel model also scored 81.2 on the AIME 2025 math test, compared to 65.4 for the older model. It also outperformed its predecessor on the MMMU-Pro Multimodal Reasoning Test, with a score of 76 compared to 69.2.

The release places particular emphasis on context management. GPT-5.5 Instant can employ the search tool to reference previous conversations, files and Gmail to provide more personalized responses. This feature will be available to Plus and Pro users on the web, with plans to roll it out to mobile devices soon. OpenAI said it plans to expand access to the feature to Free, Go Business and Enterprise users in the coming weeks.

With this update, ChatGPT will also show memory sources on all models to lend a hand you understand where it generated responses. Users can remove old-fashioned sources or correct them if the answer was wrong. Most importantly, the company stated that if you share your chat with someone, they will not be able to see your storage sources.

For developers, GPT-5.5 will be available via the API as “latest in chat”, and version 5.3 will be available as an option for paid users for three months only.

The company has faced pushback over previous model recall moves. When OpenAI retired its GPT-4o model, there was significant backlash from users who related to the “personality” of the model. GPT-4o often validated users’ choices, which made them feel connected to that particular model. Users who signed petitions to stop OpenAI from backing down, he described the model as a “best friend” or “mirror”. Despite protests, GPT-4o was retired in February 2026.

Image model updates are driving AI mobile app growth, generating 6.5 times more downloads than customary model updates, according to a novel report from app intelligence provider Appfigures.

This marks a change from earlier days when the release of novel conversation-enabled models resulted in greater demand, along with novel features such as a voice chat interface.

For example, ChatGPT and Gemini added tens of millions of novel downloads after releasing their image models, Applications found.

In the case of Google Gemini, the launch of the Nano Banana image model resulted in over 22 million downloads within 28 days of the launch of the Gemini 2.5 Flash image model last August. The data showed that after the app was introduced, the number of app downloads more than quadrupled during this period.

Meanwhile, ChatGPT added over 12 million incremental installs within 28 days of rolling out the GPT-4o image model in March last year. That’s about 4.5 times more downloads than the GPT-4o, GPT-4.5 and GPT-5 model releases, Appfigures noted.

Other model releases followed similar trends, although on a smaller scale. Meta AI’s introduction of the AI ​​Vibes video channel resulted in an estimated 2.6 million additional downloads within 28 days of its release in September 2025. (Yes, technically it’s a video model, but ultimately it’s about visual content, not just text.)

However, the report cautioned that additional downloads do not always translate into increased revenue from mobile devices.

Instead, novel versions of image models give people a reason to install the app and try out its improved image generation capabilities. However, this does not mean that they will necessarily convert to paid subscribers. For example, Appfigures noted that Nano Banana generated only $181,000 in estimated gross consumer spending within 28 days of launch, even though it drove a larger enhance in downloads than ChatGPT’s version of the 4o image model.

Meta AI’s launch of Vibes also resulted in additional downloads, but did not result in significant revenue.

Of these three, only ChatGPT converted increased attention into actual dollars.

Appfigures says OpenAI’s 4o image generation model led to estimated gross consumer spending of $70 million in the 28 days after its launch compared to its earlier baseline model.

The company also included DeepSeek in its analysis, but it did not fit the pattern.

While DeepSeek R1 was downloaded 28 million times upon its January 2025 launch, it was not a typical model comparison event. This was a watershed moment for DeepSeek, when it went from relatively unknown to an overnight sensation as the tech industry learned about the techniques it used to train its AI models at a fraction of the cost of its competitors. This case shows how curiosity can drive downloads – although in this case the interest was not linked to the image model.

According to OpenAI on Thursday, India has become the largest user base of ChatGPT Images 2.0 since its launch last week. However, third-party data reviewed by TechCrunch shows a more measured global response, with subdued overall growth and piercing spikes in select emerging markets.

ChatGPT Images 2.0, OpenAI’s latest image generation update, is designed to support more elaborate tooltips and create detailed visuals, including true text in multiple languages. The company’s early patterns suggest that users – especially in India, its largest market – are using it to create personal visuals such as avatars, stylized portraits and fantasy-themed images.

Data shared with TechCrunch by Sensor Tower and Likeweb suggests that the rollout has led to a more mixed response around the world. ChatGPT app downloads increased 11% week over week after launch, according to Sensor Tower, but overall engagement growth was modest, with daily energetic users and sessions only increasing by about 1%. Similar network data also shows confined growth in ChatGPT’s global web traffic, which was up around 1.6% week-on-week over the same period.

However, Sensor Tower data shows that some emerging markets – including Pakistan, Vietnam and Indonesia – have seen a surge in ChatGPT app downloads, up to 79% week-on-week during the rollout period.

Meanwhile, India remained the main source of activity during the rollout. Sensor Tower estimates show that ChatGPT was downloaded about 5 million times in India in its launch week, compared with about 2 million in the US, although week-over-week growth remained moderate. Data from Similar Network also shows confined growth in engagement, with daily energetic users in India growing by around 3.4% week-on-week over the same period.

In India, early trends suggest that ChatGPT 2.0 images are largely used as a form of self-expression. Instead of purely functional results, users create studio-style portraits from everyday photos, social media-ready images, and imaginative visualizations that put them in the spotlight, OpenAI says.

Early patterns also highlight differences in the adoption of AI imaging tools across markets. While India’s enormous user base is driving overall scale, sharper spikes in countries like Pakistan and Indonesia point to greater demand from novel users in emerging markets post-launch.

The launch of OpenAI Images 2.0 comes amid growing competition in AI image generation. Google’s earlier image-centric model, the original Nano Banana, also saw robust early adoption in India, indicating that the country has become an significant market for image generation.

With the novel version of ChatGPT Images, OpenAI goes further with improvements such as better rendering of text in non-Latin languages, including Hindi and Bengali, and novel “thinking” capabilities that allow it to refine results and generate multiple variations from a single prompt.

OpenAI says that beyond stylized portraits and avatars, early adopters of Images 2.0 in India are experimenting with a wider range of formats – from fantasy newspaper covers to tarot-style graphics and fashion mood boards. Users also operate the AI ​​tool to restore older photos and create cinematic portrait collages, the company says, suggesting early patterns of more personal operate.