AI News

Same-Day Moves: DeepMind Gives 10,000 Scientists Free Access to Gemini 4 Carbon as OpenAI Expands to 200,000 Researchers

On November 16, 2026, Google DeepMind and OpenAI announced parallel expansions of free access programs for researchers, with DeepMind offering Gemini 4 Carbon to up to 10,000 scientists and OpenAI raising its academic researcher program cap to 200,000. The moves intensify competition over scientific workflows, pricing benchmarks, and real-world validation data.

Google DeepMind OpenAI Gemini 4 Carbon
31

Trump Plans to Designate Over 25 U.S. AI Firms as National Champions: Chip Controls Give Way to Champion-Backed Global Expansion

The Trump administration is preparing to designate more than 25 U.S. AI companies, including Nvidia, OpenAI, Amazon, Microsoft and Anthropic, as “national champions,” pairing the designation with fast-track export approvals, government backing for overseas expansion and priority access to federal financing. The move marks a pivot from restrictive chip controls toward actively packaging and promoting U.S. AI capabilities abroad.

特朗普 AI政策 芯片出口
33

OpenAI Healthcare Agent Launches November 16; CVS Clears Months of Backlog Over a Weekend; 99% Safety Rate Faces Compliance Scrutiny

OpenAI has launched a major ChatGPT for Healthcare upgrade with computer-use agents powered by GPT-6 Astra and GPT-6 Luna, and CVS Health used them to clear months of document backlog in a single weekend. The 99% clinical safety rate achieved by GPT-6 Astra now faces compliance scrutiny as healthcare deployments move into high-compliance workflows.

AI医疗 OpenAI 临床代理
60

Tencent Releases Hy4 Flash: 224B Parameters, 1M Context, Cost Just 1/18 of Claude Haiku 5.5—But Key Data Comes from Internal Tests

Tencent has released Hy4 Flash, a 224B-parameter model with native multimodality and a 1M-token context, priced at $0.070 per million input tokens and $0.281 per million output tokens. Tencent claims near-Claude Haiku 5.5 user satisfaction at roughly 1/18 the cost per task, but all performance comparisons come from internal tests and have not yet been independently reproduced.

腾讯混元 大模型 AI定价
96

Google First Reports Russian-Language Hacker AI Framework Achieving Fully Automated Exploit Deployment, Unattended Attacks on French Media and Education Sites

Google's Threat Intelligence team disclosed that Russian-speaking attackers used an AI framework called exploit_agent to complete the full chain from patch diff scanning and vulnerability identification to exploit code generation and automated attack deployment, with no human intervention, targeting French media, entertainment, and education websites.

AI Safety 网络攻击 漏洞利用
121

OpenAI Opens Training Internals to Four Independent Organizations: Six Researchers Embedded to Assess AI Subversive Behavior

OpenAI has signed agreements with Apollo Research, METR, Transluce, and Goodfire to launch what it calls a “Phase 1 embedded evaluation,” with Apollo embedding six evaluators for four months to investigate whether AI models might actively subvert safety mechanisms. The arrangement moves external safety review into the training process, but access and disclosure limits remain unresolved.

OpenAI AI Safety Apollo Research
180

US Closes Indirect AI Compute Loophole: Commerce Department Moves to Control Remote Access After Jensen Huang Resigns

The US Commerce Department issued an interim export control rule requiring licenses before remote AI compute access is provided to restricted countries when a data center or compute network exceeds 12 million TPP. The measure targets offshore data centers used as intermediaries and follows scrutiny of Megaspeed and Jensen Huang's resignation.

AI出口管制 NVIDIA Jensen Huang
204