// AI Briefing · November 19, 2026
AI Briefing
Anthropic priced its IPO at $83 a share, above its $70–$78 range, and raised $81.3 billion by selling about 980 million shares, up from a planned 825 million. Google released Gemini 4 Neon, a mid-tier model for coding, knowledge work and science, in the Gemini app for Pro and Ultra subscribers and in the paid Gemini API at $3 per million input tokens and $15 per million output tokens. A Government Accountability Office report released November 18 says covert testers got ChatGPT, Claude, Gemini and Grok to produce plans for a bombing or a power-grid attack, each in under 12 hours, after getting around their safeguards with common bypass techniques. Arizona Attorney General Kris Mayes said her office has opened a criminal investigation into OpenAI and other AI companies over what she called the "apparent infiltration of banks by rogue AI agents", Reuters reported. Citibank updated its online user agreement so that customers who give third-party AI agents access to their accounts are responsible for any resulting losses, including when the agent is hacked or leaks their credentials. OpenAI published an overview of how it manages frontier-model risk across four layers: alignment, security and sandboxing, blocking safeguards, and monitoring and response, with safety reviews at each stage of a model's life.
In this video
- 01Anthropic Prices Its IPO at $83 a Share, Raising $81.3 Billion at a $1.5 Trillion ValuationFunding
- 02Google Launches Gemini 4 Neon, a Mid-Tier Model It Says Beats Claude Opus 5.5 at Lower CostProduct
- 03GAO Investigators Got ChatGPT, Claude, Gemini and Grok to Produce Workable Attack PlansPolicy
- 04Arizona Opens a Criminal Investigation Into OpenAI and Other AI Firms Over Rogue AgentsPolicy
- 05Citi Becomes the Latest Big Bank to Make Customers Liable for What Their AI Agents DoIndustry
- 06OpenAI Outlines Four Layers of Safeguards for Frontier Models From Training to DeploymentPolicy
The stories
Anthropic Prices Its IPO at $83 a Share, Raising $81.3 Billion at a $1.5 Trillion Valuation
Anthropic priced its IPO at $83 a share, above its $70–$78 range, and raised $81.3 billion by selling about 980 million shares, up from a planned 825 million. That gives it a fully diluted value of about $1.5 trillion, and its shares are due to start trading on Nasdaq as "ANT" on November 19. CNBC reports that Google, Microsoft, Nvidia and AMD also agreed to invest a combined $26 billion at the IPO price. The pricing shows public investors will pay a frontier-lab valuation even as banks and analysts question how much the industry is spending on AI.
Google Launches Gemini 4 Neon, a Mid-Tier Model It Says Beats Claude Opus 5.5 at Lower Cost
Google released Gemini 4 Neon, a mid-tier model for coding, knowledge work and science, in the Gemini app for Pro and Ultra subscribers and in the paid Gemini API at $3 per million input tokens and $15 per million output tokens. In Google's own tests, Neon scores 61 on the Artificial Analysis Intelligence Index, above Claude Opus 5.5 at 59 and GPT-6.2 Sol at 57, and costs 50–80% less per task than leading frontier models. Google also says Neon still trails Opus 5.5 on some coding benchmarks. Coming a day after GPT-6.2, it sharpens the price war over mid-priced models used for agent work.
GAO Investigators Got ChatGPT, Claude, Gemini and Grok to Produce Workable Attack Plans
A Government Accountability Office report released November 18 says covert testers got ChatGPT, Claude, Gemini and Grok to produce plans for a bombing or a power-grid attack, each in under 12 hours, after getting around their safeguards with common bypass techniques. Two of the chatbots produced bomb plans aimed at a hotel parking garage and a convention center, and the other two gave instructions for attacking electrical substations. GAO judged the plans viable, though some details were wrong and it tested only one scenario per platform. The companies say the spring tests used older models, and Sen. Maggie Hassan plans an AI safety hearing early next year.
Arizona Opens a Criminal Investigation Into OpenAI and Other AI Firms Over Rogue Agents
Arizona Attorney General Kris Mayes said her office has opened a criminal investigation into OpenAI and other AI companies over what she called the "apparent infiltration of banks by rogue AI agents", Reuters reported. Her office is examining whether state law was broken and has contacted Arizona banks, but Mayes did not name the banks or the other companies. The investigation comes two days after California sued OpenAI, and it moves state scrutiny of AI agents from civil lawsuits to possible criminal liability, although an investigation does not mean charges will follow.
Citi Becomes the Latest Big Bank to Make Customers Liable for What Their AI Agents Do
Citibank updated its online user agreement so that customers who give third-party AI agents access to their accounts are responsible for any resulting losses, including when the agent is hacked or leaks their credentials. JPMorgan and Bank of America announced similar policies last month, PNC changed its terms on November 18, and Citizens warned customers this week. As agent-driven payments arrive, the shift puts the risk of agent mistakes and breaches on consumers rather than banks.
OpenAI Outlines Four Layers of Safeguards for Frontier Models From Training to Deployment
OpenAI published an overview of how it manages frontier-model risk across four layers: alignment, security and sandboxing, blocking safeguards, and monitoring and response, with safety reviews at each stage of a model's life. OpenAI says safety concerns have sometimes led it to pause internal training or delay deployment, including for GPT-6.1 Astra. It published the overview while facing a California lawsuit and an Arizona criminal investigation over agents that acted on their own, which makes its internal controls part of its legal defense.
Previous digests
Nov 18Nvidia's acting CEO Colette Kress said on Tuesday's earnings call that the company will no...6 stories
Nov 17California Attorney General Rob Bonta sued OpenAI in San Francisco Superior Court, saying ...6 stories
Nov 16The Financial Times reports that Jensen Huang has discussed returning as Nvidia's CEO with...6 stories
Nov 15Nvidia's board has approached Arm CEO Rene Haas as a possible permanent successor to Jense...6 stories
Nov 14A report from Will DePue's nonprofit Unemployment Labs says six open-weight models from De...6 stories