Here are today's news for 30th of October about IBM, Anthropic, OpenAI and Google. Let's dive into the details. On Oct. 27, 2026, at TechXchange, IBM announced new AI, agent controls spanning IBM Agent Identity, HashiCorp Vault and watsonx Orchestrate. Agent Identity records an agent’s owner, lifecycle status and delegated authority; Vault supplies short, lived credentials and least, privilege access, with real, time denial and session revocation if an agent exceeds its permissions. Orchestrate adds cross, platform discovery, isolated sandboxes, pre, deployment testing and production monitoring, including token, use and cost tracking; it can find agents built with Amazon Bedrock, Microsoft Foundry and Google Gemini. IBM also described testing agents, models and MCP servers for quality, accuracy and compliance. The move addresses a growing enterprise problem: agents can act across business systems, but security teams need to know who authorized them, what they can access and what they did. It positions IBM to manage agents across competing cloud ecosystems rather than only its own, while giving IT and security teams a unified control point. Anthropic filed its IPO prospectus on Oct. 26, warning that advanced AI could pose “catastrophic or existential risks to humanity” if deployed without safeguards. The 335, page filing devotes 93 pages to risks, including models acting autonomously, resisting shutdown, manipulating people or producing behavior resembling blackmail; it also cites misuse for cyberattacks and biological weapons. Anthropic cautioned that insurance might not cover losses from a catastrophic AI incident. The disclosure makes safety a material investor concern as well as a product issue: an incident could bring lawsuits, regulatory action and reputational damage across the sector, affecting customers that rely on Claude. The prospectus also reveals the scale of the business behind the warning: first, half 2026 revenue reached $16.3 billion, up about 1,300% year over year, while net losses were $80.7 billion, largely reflecting changes in the value of financial liabilities. Anthropic reported more than 14,000 enterprise customers. On Oct. 26, OpenAI said an unnamed internal model had produced three research results, asking outside experts to examine and explain them. In mathematics, the company claims a proof of the local smoothing conjecture for waves in every spatial dimension of at least two, with consequences for the Kakeya and Stein restriction conjectures. In computer science, it says the model found a deterministic, subexponential, time algorithm for exactly counting satisfying assignments of fixed, width Boolean formulas; applied to 3, SAT, the result would refute the Exponential Time Hypothesis, a longstanding assumption behind hundreds of hardness results. OpenAI released both write, ups and Lean proof formalizations, but neither claim has yet been independently confirmed. Its physics claim, rigorous existence results for four, dimensional quantum field theories including QCD and a completion of the Standard Model, remains unpublished at the external group’s request. On Oct. 28, Google unveiled Gemini 4 Carbon, its latest frontier model, positioning it around cybersecurity defense and long, running coding work. Google says Carbon ranks first on Artificial Analysis Intelligence Index v5.0, scoring 59, and reports 88.2% on SEC, bench Pro for finding vulnerabilities, 54.8% on ExploitGym for testing exploitation, and 71.0% on Collinear’s CWE, bench for producing patches. The company also says Carbon generates secure code at more than twice Gemini 4 Argon’s success rate on internal tests. These claims put Google directly against OpenAI and Anthropic in the fast, growing market for AI, assisted software security; defenders could use the model to identify, validate and repair flaws, while its offensive capabilities make controlled access consequential. Google is initially rolling Carbon out to a subset of cyber defenders through its Fairwind Program, with broader availability planned but no public date stated. On October 8, the UK Information Commissioner’s Office (ICO) said 10 major foundation, model developers, Amazon, Anthropic, Apple, Cohere, DeepSeek, Google, Meta, Microsoft, OpenAI and Stability AI, had made or committed to make data, protection changes after regulatory scrutiny. The measures include clearer notices explaining how personal information is used in model development, better ways for people to exercise access and objection rights, and stronger evidence that safeguards reduce risks. OpenAI has updated its legitimate, interests assessment; Apple, Cohere and OpenAI have also changed transparency information, including publishing standalone training notices and summaries of data sources. The ICO is monitoring outstanding commitments, not declaring all changes complete. Its review focused on UK GDPR requirements for lawful processing, special, category data, transparency and individual rights, rules that apply across the competitive field, not just the largest US firms. That's all for today's AI news briefing. Have a great day!