News, p. 35
News
440 stories · page 35 of 49OpenAI proposes the framework by which it wants to be measured
On 15 July OpenAI laid out a proposal for regulating frontier AI model safety - a mix of state laws, a strengthened federal evaluation body, and a resilience plan against the worst risks. The company that builds the most powerful models is writing the rules by which it wants to be judged. Just sit with that.
Read →NVIDIA brings its robot chips down to the mass market
On 15 July NVIDIA launched two new Jetson Thor modules - T3000 and T2000 - on the Blackwell architecture, for robotics and edge AI. The cheaper one lands exactly where earlier robot boards were too expensive for a device made in volume.
Read →Quantum team achieves universal operations by braiding particles
On 15 July a team from Quantinuum, Harvard and partners published in Nature a complete set of universal quantum operations obtained by braiding and fusing anyons on a real quantum processor. A step toward qubits where error protection is built into the physics itself, not patched on top.
Read →VS Code moves its AI-agents into a separate process
Version 1.129 introduces 'Agent Host' - a separate process that runs agent harnesses (Copilot, Claude, Codex), so one session can run across several windows and you can manage many agents at once. Plus a redesigned panel that brings review and chat together in one place.
Read →xAI sues a user for bypassing Grok's safeguards
xAI filed a civil suit against a man from South Carolina, whom it accuses of opening accounts under a false identity and constructing deceptive prompts to bypass Grok's safeguards and produce sexualized images of minors. The company is seeking damages and a permanent ban from access.
Read →OpenAI launches an automated attacker that hardens its models
OpenAI unveiled GPT-Red - an internal tool that automatically attacks its own models (red-teaming) and feeds what it learns back into their training. Per the company, the new version achieves noticeably fewer breaches under prompt injection than its predecessor from months ago.
Read →Thinking Machines released Inkling - an open model they don't even market as the strongest
Mira Murati's lab unveiled its first open-weights model. The interesting part is the bet: a base you can download and fine-tune for yourself.
Read →Thinking Machines open-sourced a frontier-class model
On 15 July the lab led by Mira Murati - OpenAI's former CTO - showed Inkling, its first public model, and released it with open weights. It's not a small demo model: a large mixture-of-experts class, multimodal, with a long context window, that anyone can download and run themselves.
Read →A critical hole in Oracle E-Business Suite joins the actively exploited list
On 15 July, CISA added a vulnerability in Oracle E-Business Suite to its KEV catalog - one that allows unauthenticated takeover of the payments module. The deadline for US federal agencies falls on 18 July. Three days. The same day, an older hole in the KNX building-automation protocol joined the list too.
Read →