https://aidaily.si/stories/xbow-says-ai-agents-built-a-chrome-exploit-chain-that-won-a-250-000-award.htmlAI Switchboarden2026-10-14T05:00:00-05:00XBOW says one researcher and OpenAI's GPT-5.6 Sol models built a Chrome exploit chain that earned Google's $250,000 award, in six days https://aidaily.si/stories/us-and-european-bank-supervisors-step-up-ai-questions-reuters-reports.htmlAI Switchboarden2026-10-14T05:00:00-05:00US bank regulators are questioning lenders about rogue AI agents and Chinese-built models, and British supervisors want “bare metal” fallbacks, Reuters reports https://aidaily.si/stories/spacexai-agrees-to-testify-to-new-york-city-council-after-subpoena.htmlAI Switchboarden2026-10-14T05:00:00-05:00SpaceXAI agrees to testify to the New York City Council on Friday, after skipping its AI safety hearing and facing a court petition over its subpoena https://aidaily.si/stories/sec-declares-agility-robotics-merger-filing-effective-vote-set-for-2-november.htmlAI Switchboarden2026-10-14T05:00:00-05:00The SEC declares Agility Robotics' merger filing effective, putting the humanoid maker's Nasdaq listing to a shareholder vote on 2 November https://aidaily.si/stories/reuters-anthropic-and-openai-sought-control-over-what-outside-evaluators-publish.htmlAI Switchboarden2026-10-14T05:00:00-05:00Anthropic and OpenAI sought control over what outside safety evaluators may publish, Reuters reports, weeks after promising them independence https://aidaily.si/stories/protoarena-ai-agents-exploit-network-protocol-flaws-in-12-5-to-48-8-of-tests.htmlAI Switchboarden2026-10-14T05:00:00-05:00An Opus 4.8 agent exploited 48.8% of network-protocol flaws when given guidance and 12.5% when working blind, a new benchmark finds https://aidaily.si/stories/physical-intelligence-opens-an-api-beta-for-its-robot-models.htmlAI Switchboarden2026-10-14T05:00:00-05:00Physical Intelligence opens a waitlisted API beta so other companies can fine-tune and run its robot models https://aidaily.si/stories/openai-will-retire-gpt-4o-and-gpt-4-1-from-its-api-on-14-january.htmlAI Switchboarden2026-10-14T05:00:00-05:00OpenAI will remove GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini from its API on 14 January 2027, and points developers to GPT-6 Luna https://aidaily.si/stories/hinton-bengio-and-415-scientists-call-for-an-international-ai-containment-body.htmlAI Switchboarden2026-10-14T05:00:00-05:00Hinton, Bengio and more than 400 scientists call for an international body to set AI containment standards, and for containment failures to be reported to it https://aidaily.si/stories/google-makes-gemini-4-argon-generally-available-at-an-introductory-price.htmlAI Switchboarden2026-10-14T05:00:00-05:00Google opens Gemini 4 Argon to all paying API developers and to Pro and Ultra subscribers, at an introductory $2 and $10 per million tokens that doubles on 1 December https://aidaily.si/stories/asml-says-planned-2027-and-2028-supply-will-not-meet-demand.htmlAI Switchboarden2026-10-14T05:00:00-05:00ASML reports €11.7bn of third-quarter sales and says its planned supply for 2027 and 2028 will not fully meet customer demand https://aidaily.si/stories/apple-launches-homeview-with-siri-ai-and-plans-a-fee-for-more-use.htmlAI Switchboarden2026-10-14T05:00:00-05:00Apple launches HomeView, a $299 home screen built around Siri AI, and says expanded access to its server-side AI will cost a fee https://aidaily.si/stories/xpeng-says-its-iron-humanoid-is-planned-for-global-rollout-in-2027.htmlAI Switchboarden2026-10-13T05:00:00-05:00XPeng's own release says its IRON humanoid robot is planned for global rollout in 2027, and TechNode reports a European launch later that year https://aidaily.si/stories/uk-aisi-proposes-eight-priorities-for-frontier-ai-best-practice.htmlAI Switchboarden2026-10-13T05:00:00-05:00The UK AI Security Institute sets out eight priorities for technical best practice at frontier labs, including reporting incidents caused by AI agents https://aidaily.si/stories/uk-ai-minister-tells-labs-do-not-release-what-they-cannot-control.htmlAI Switchboarden2026-10-13T05:00:00-05:00Britain's AI minister tells frontier labs “do not release the system” if they cannot control its risks, and says a breach of UK cyber defences seems likely https://aidaily.si/stories/tencent-tells-bond-investors-its-5bn-offering-will-mainly-fund-ai-spending.htmlAI Switchboarden2026-10-13T05:00:00-05:00Tencent tells bond investors its roughly $5bn offering is meant to fund AI spending, mostly on chips, Bloomberg reports https://aidaily.si/stories/south-korea-clears-20-robros-humanoids-to-work-beside-people.htmlAI Switchboarden2026-10-13T05:00:00-05:00South Korea clears 20 walking humanoid robots from Robros to work beside people at chip, car, shipyard and warehouse sites https://aidaily.si/stories/meta-releases-muse-wonder-its-first-frontier-model.htmlAI Switchboarden2026-10-13T05:00:00-05:00Meta releases Muse Wonder, its first frontier model, at $4 per million input tokens and $20 per million output tokens https://aidaily.si/stories/kospi-falls-5-46-as-sk-hynix-drops-10-on-cxmt-worries.htmlAI Switchboarden2026-10-13T05:00:00-05:00Seoul's KOSPI falls another 5.46% and SK hynix drops 10%, as Korean and AP reports point to China's memory maker CXMT https://aidaily.si/stories/intel-and-broadcom-ceos-say-capital-may-be-ai-s-biggest-constraint.htmlAI Switchboarden2026-10-13T05:00:00-05:00The chief executives of Intel and Broadcom tell a Singapore conference that capital, not chips, may be AI's biggest constraint https://aidaily.si/stories/brad-smith-says-a-washing-machine-faces-more-safety-rules-than-ai.htmlAI Switchboarden2026-10-13T05:00:00-05:00Microsoft's Brad Smith says a washing machine faces more safety rules than AI, and calls for outside testers before release