Full Summary
This Thursday morning, OpenAI and Broadcom have unveiled "Jalapeño," their first custom AI chip, developed in a record nine months. Both 헤럴드경제 and Verdict.co.uk confirm this chip aims to rival Nvidia's Blackwell and Google's TPUs, with early tests showing superior performance per watt. OpenAI plans to deploy Jalapeño in data centers later this year, reducing its reliance on Nvidia and potentially making AI faster and more affordable. Meanwhile, Anthropic is facing intense pressure on multiple fronts. The BBC and Akses.co.id report that Anthropic accuses Alibaba of "industrial scale" AI theft, alleging nearly 29 million exchanges with its Claude AI model using thousands of fraudulent accounts to "distill" its capabilities. On the domestic side, NewsBytes and the New York Post reveal Anthropic CEO Dario Amodei was replaced in high-stakes White House negotiations by co-founder Tom Brown, due to Amodei being perceived as "difficult to work with." These talks aim to lift an export control order on Anthropic's Fable 5 and Mythos 5 AI models, which Cybernews reports led to a lawsuit against the US government by a legal tech company. In other AI news, nearly 400 newspapers are suing Microsoft and OpenAI, claiming their AI models illegally scraped copyrighted content, as reported by Shacknews. And the US government, according to Yahoo and Let's Data Science, has asked OpenAI to stagger the release of its new GPT-5.6 model due to security concerns, with access to be approved customer by customer. This convergence of events means your access to the latest AI tools could be directly impacted by legal battles, international espionage, and shifting government regulations, potentially slowing down innovation or increasing costs.