Welcome to the newest edition of ID Tech’s AI update. Here’s the latest big news on the shifting landscape of AI and identity technology:
The White House is reportedly preparing new AI chip restrictions with respect to Malaysia and Thailand due to ongoing concerns about a suspected black market feeding chips to China despite sanctions. A draft Commerce Department rule would prevent firms from obtaining advanced AI chip components from intermediaries located in the targeted countries.
The US Senate has voted overwhelmingly against a plan to scrap state-level AI regulations in the Trump administration’s “Big Beautiful Bill”. An amendment to the bill, approved in a 99-1 vote, retains state powers over matters including deepfake safety, autonomous vehicles, and online child protections. The “Big Beautiful Bill” has passed with the amendment.
The European Commission has flatly rejected a last-minute plea from European tech titans urging that the EU Artificial Intelligence Act be pushed back two years. The CEOs of Airbus, chip-equipment giant ASML, and French LLM maker Mistral AI argued in an open letter that the complex and overlapping requirements of the regulatory package would hurt Europe’s competitiveness in the AI space. But the EC said the law’s deadlines are, at this point, legally binding.
California-based Dexterity has reached a valuation of $1.65 billion after a $95 million funding round. The round was co-led by returning backer Lightspeed Venture Partners and new strategic investor Sumitomo Corporation, and will further fuel the firm’s development of “superhumanoid” robots called “Mechs” that Dexterity has been building in collaboration with Kawasaki Heavy Industries. The Mechs are already being trialled by FedEx, UPS, GXO, and Sumitomo.
Genesis AI emerged from stealth this week with a $105 million seed round jointly led by Khosla Ventures and Eclipse. The AI firm, whose offices are split between Paris and Palo Alto, is working on a universal foundation model capable of controlling any robot. The startup’s 20-person team is drawn from some big names in the space, including CMU, MIT, Nvidia, and Mistral AI.
OpenAI and Oracle have massively expanded their “Stargate” partnership, with the latter agreeing to provide about 4.5 GW of extra U.S. data center power in exchange for about $30 billion a year starting in Oracle’s fiscal 2028. Oracle will double its existing Abilene, Texas “supercluster” to 2 GW and is scouting additional sites in Michigan, Wisconsin, Wyoming, New Mexico, Georgia, Ohio, and Pennsylvania to meet demand. Their $500 billion “Stargate” venture, in partnership with SoftBank, was announced earlier this year.
Meta is in talks to raise $29 billion in credit to help fuel the construction of new AI datacenters in the US. Apollo, KKR, Brookfield, Carlyle, and Pimco are reportedly interested, and the deal would entail $3 billion in equity plus $26 billion of long-dated debt arranged by Morgan Stanley. Lenders would get a direct stake in the facilities and issue tradeable notes.
Apple has quietly approached OpenAI and Anthropic about using their models to power the AI on its own devices, according to a new report. It’s a tacit admission that Apple has fallen way behind rivals in its own internal AI efforts; now, it needs to turn to others to help it catch up. Early tests reportedly show Anthropic’s Claude outperforming Apple’s own model – no surprise there – but also Google’s Gemini, which is generally considered state-of-the-art.
Akool has unveiled a new AI tool that is designed to create photorealistic avatars for users that can reflect their expressions, gestures, and tone, in any language, in real time. That means another avatar, which could look like someone else, repeats what you say and how you say it, potentially in a different language, as you’re speaking. In a social media post announcing Akool Live Camera, the firm called it “the future of video, where anyone, anywhere, can show up fully and authentically”.
The chatbot’s take: One for the NGE fans out there.
–
By Alex Perala
