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:
OpenAI has tightened its internal security amid concerns about corporate espionage. The company has reportedly implemented fingerprint scanning for physical access to sensitive areas, and an “information-tenting” policy that strictly limits which employees can view or discuss confidential projects. The anxiety stems at least in part from concerns that China-based DeepSeek attempted to copy its models.
Nvidia has become the world’s first $4 trillion company in market value, thanks to ongoing, AI-driven demand for its Graphics Processing Units (GPUs). The company’s stock is up about 22 percent in 2025 and more than 870 percent since the ChatGPT-sparked AI boom began in late 2022.
Nvidia rival Groq is in talks to raise $300 million to $500 million in a new funding round that could value it at about $6 billion. The firm specializes in application-specific integrated circuits (ASICs) called Language Processing Units (LPUs), engineered specifically for running large language models (LLMs) and other AI tasks much faster and more efficiently than general-purpose GPUs.
China-based Zhipu AI is considering an Initial Public Offering on Shanghai’s STAR Market. The firm says its GLM-4 model beats GPT-4 on some benchmarks, and is looking to capitalize on its relatively strong position as a domestic AI firm. But mainland regulators have slowed approvals for high-profile tech floats, while Hong Kong’s deal flow is rebounding.
Meta has wooed another top tech engineer with a pay package worth more than $200 million over a multi-year period, according to new reports. Ruoming Pang ran Apple’s 100-person Foundation Models group that built the LLMs behind new “Apple Intelligence” features. The news comes after OpenAI CEO Sam Altman made comments about Meta offering $100 million signing bonuses in its efforts to poach talent.
Meta has acquired a stake of just under three percent in EssilorLuxottica SA, the owner of Ray-Ban and Oakley. The aim is to get privileged access to the world’s biggest eyewear manufacturing and retail network as Meta seeks to develop its AI smart glasses pipeline. The transaction cost about $3.5 billion at current market prices, and Meta has the option to raise its holdings to about five percent over time.
CoreWeave will acquire Core Scientific, a publicly-traded bitcoin-miner-turned-data-centre-operator, in a deal valued at about $9 billion. The move comes shortly after CoreWeave’s IPO on the Nasdaq in March. Core Scientific brings to the table about 1.3 GW of contracted power across U.S. sites, plus more than 1 GW of expansion headroom, which will help CoreWeave to meet its own future lease obligations with respect to data centers.
The European Commission has published a three-chapter, voluntary “GPAI Code of Practice” covering Transparency, Copyright and Safety & Security for developers of general-purpose AI models. It’s meant to help companies comply with the impending AI Act. While the Code is voluntary, the underlying AI Act starts to bite August 2nd for new GPAI models and exactly two years after that for existing ones, with fines up to seven percent of global turnover for non-compliance.
xAI officially launched Grok 4, an LLM that has reached a new state-of-the-art with respect to at least some testing benchmarks. It has about 1.7 trillion parameters and an expanded 256 K-token context window, and CEO Elon Musk says it’s “the smartest AI in the world.” Unfortunately, the launch came just hours after Grok 3 began posting antisemitic replies that lauded Adolf Hitler and recycled extremist stereotypes on Musk’s social media platform – behaviour that Musk attributed to flawed training data.
The chatbot’s take: We asked for a bit more colour on Meta’s latest poach, which seemed at first like a weird move.
–
By Alex Perala
