OpenAI has cut its fundraising target from $30 billion to a maximum of $20 billion. The round is still open, and the Financial Times reported that the company is seeking a $1.25 trillion valuation.
Google has joined OpenAI, Anthropic and Microsoft in pledging to honor a shared website setting that lets publishers block future scraping for AI training. The setting will not remove content already collected or keep pages out of search results.
S&P Global Ratings changed Meta's credit outlook from stable to negative, citing heavy AI spending and negative free cash flow. Meta shares fell 15% in October, their worst month in four years.
Researchers at parameter.ai showed that opening a malicious shared ChatGPT conversation could let an attacker read a user's private chats and act through their account. OpenAI confirmed the flaw and fixed it on October 26.
Users report that Anthropic's Claude AI is refusing requests to recreate existing games and, in some cases, help reverse-engineer software. Some developers say the refusals have interrupted legitimate work, but Anthropic has not confirmed the rules or explained the decisions.
Apple CEO John Ternus says the company will use AI when it makes products better, rather than add it for show. He says Tim Cook can help him with policy matters when asked, and he has not met President Donald Trump.
Claude Duet is in a limited US preview for Claude Pro and Max subscribers, who must join a waitlist before they can use it. Once admitted, users can connect apps and ask it to handle tasks such as scheduling meetings or preparing daily briefings.
Google Deepmind and the World Resources Institute built ForTy, a worldwide map that uses satellite images and AI to identify different types of forest and tree farms in 2020. It was accurate in 90.2% of test cases overall, but often missed planted forests and tree crops.
Microsoft is offering voluntary buyouts to about 60,000 employees across four countries, or roughly a quarter of its staff. The offer covers Xbox, Windows and Copilot workers but excludes teams such as data centers and cybersecurity that support its AI plans.
Cisco Talos reports that VOLTA malware steals passwords, account sessions and other personal data, then uses large language models to rank victims’ assets by potential value. If one model refuses the task, the malware can try others.