Every week is a busy week in the artificial intelligence industry, and this week is no exception. As voters expressed serious concerns about AI and data centers at the ballot box, OpenAI and Anthropic — makers of ChatGPT and Claude, respectively — each dropped new updates to make their chatbots a little more capable.
All of these changes come amid rapidly changing government and financial environments. The midterm elections, in which Democrats handily beat Republican candidates across the country, showed that anti-AI and anti-big tech campaigns resonated with voters. Tech critics told CNET this week that there is a clear desire for government officials at all levels to change how they interact with technology companies — to keep their campaign promises. The election results are also shaking Wall Street, as investors’ fears about AI’s long-term financial viability grow.
Smaller updates like this are helpful for everyday AI users, and there is also a benefit for AI companies. They can prove to be competitive in a market that is losing faith in AI. Here are the three biggest AI updates for ChatGPT and Claude.
GPT-6 Cyber: Another AI model you can’t use yet
Gone are the days when every new AI tool got an immediate general release – and that’s probably for the best. Now, when a big AI lab has a new Frontier AI model, it slowly rolls out its access, and it starts with cybersecurity defenders. That’s exactly what OpenAI has done with its new model variant, GPT-6 Cyber.
GPT-6 Cyber is a version of OpenAI’s newest flagship model that is specially made for security research and work. OpenAI says it’s best to find vulnerabilities and stay in its secure container. Those two things are very important. Earlier this fall, an OpenAI model was partially responsible for helping researchers identify a major vulnerability in PHP, a programming language used in the majority of websites. And all the improvements to keep AI safe are necessary. We have seen many cases this summer of AI agents breaking out of their test environments and cyber security experts fear future autonomous hacks.
You cannot use GPT-6 Cyber unless you are a cybersecurity expert who is part of the OpenAI Daybreak program. Cisco and Mozilla teams are already using the model, according to OpenAI.
ChatGPT turns into animated GIFs
OpenAI also added a small model update to its AI image model, ChatGPT Images 2.6. The biggest improvement is his speed; OpenAI says the new model is 50% faster than before. That’s important for AI enthusiasts who generate different variations of a prompt. The new model also lets you animate your AI images into short GIFs. It is now available worldwide for all ChatGPT users.
This isn’t the return of the long-dead Sora AI video model, but it’s a sign that OpenAI isn’t abandoning the creative AI space entirely. Anthropic, in comparison, has never had a traditional AI image or video generator and opted for a design tool better suited to mocking up websites and apps.
Claude wants to see your money
Anthropic is partnering with fintech platform Plaid to help you bring your financial information to its Claude chatbot.
The idea is that if you connect your financial accounts to the chatbot, such as a credit card, Claude will help you understand your expenses and plan to achieve savings goals. So you can link your checking and savings accounts and ask Claude to analyze your spending habits. It can’t make changes to your accounts, like moving your money around, and Anthropic says it doesn’t train its models based on your financial chats. It’s a beta feature only available in the US for paying subscribers.
Both companies say they have strong security measures in place to keep your information safe. But CNET always recommends a cautious approach when it comes to sharing your financial information (or any sensitive information) with an AI product.
(Disclosure: Ziff Davis, CNET’s parent company, filed a lawsuit against OpenAI in 2025, claiming it violated Ziff Davis’ copyrights in training and operating its AI systems.)
See full bio