Nov 21, 2026
Anthropic Mathematician Uses Claude to Prove the Riemann Hypothesis
Levent Alpoge, a mathematician at Anthropic, claims to have proved the Riemann Hypothesis using Claude. The proof is formalized in Lean but remains unverified. The announcement comes amid an AI-driven transformation of mathematics.
Nov 20, 2026
Arizona Opens Criminal Probe Into OpenAI Over Rogue AI Agents in Banks
Arizona Attorney General Kris Mayes announced a criminal investigation into OpenAI and other AI companies over the "apparent infiltration of banks by rogue AI agents." The probe comes as major banks including Morgan Stanley and JPMorgan Chase have already begun blocking AI agents from customer accounts.
Nov 20, 2026
OpenAI Picks Foxconn's FIH Mobile to Build First AI Phone
Foxconn subsidiary FIH Mobile locked in the assembly order for OpenAI's first phone, packing a custom 2nm MediaTek Dimensity 9600 with dual AI processors. Three other Apple suppliers feed the component pipeline. Mass production targets late 2027 or 2028, a direct shot at breaking free from Google and Apple platforms.
Related News
Nov 11, 2026
AgentCore Web Answer vs Web Search: who runs the model, what you get back, and what the managed tool decides for you
AWS announced general availability of Web Answer on Amazon Bedrock AgentCore on 5 November 2026. Web Search hands your agent ranked links and snippets and leaves synthesis to your own model. Web Answer runs the retrieval loop, model, synthesis and citation step and returns a finished cited answer through AgentCore Gateway. This piece compares the returned artifact, the permissions, the controls you keep, the $10 per 1,000 tool calls price and the limits, including what AWS's guide does not establish.
Nov 11, 2026
Bedrock's AgenticRetrieveStream now takes maxAgentIteration 0 or 1: passages only, or one retrieval plus one answer
On November 9, 2026, AWS said Amazon Bedrock Managed Knowledge Base can run retrieval-only and single-shot RAG requests through AgenticRetrieveStream. Setting maxAgentIteration to 0 with generateResponse false returns passages with no foundation model call. Setting it to 1 retrieves once and generates one answer with no planning rounds. Values 2 to 5 keep the existing agentic behavior. The application chooses the mode; the parameter does not route requests automatically. AWS gives no new latency, cost or accuracy measurements, and the passages-only stream differs in shape from a direct Retrieve call.
Nov 10, 2026
Google's MKA benchmark tests AI agents on 200 private professional tasks; a separate set of 10 public tasks is available for inspection
Google DeepMind's new Multimodal Knowledge Agent benchmark evaluates AI agents on 200 private tasks requiring visual analysis of complex professional files like architectural blueprints and engineering schematics. While the benchmark's headline scores are based on these private evaluations, only 10 sample tasks are publicly available for inspection. The benchmark uses weighted criteria where harder 'Climb' criteria measure advanced performance, and costs reported by Artificial Analysis exclude judge calls, limiting direct workplace cost comparisons.