AI News for October 6, 2026 — 7min.ai

[NEWS] Reflection AI releases Beam, a 501B open-weight model to rival China's best: Reflection AI unveiled Beam, its first frontier open-weight model, built for coding, reasoning and agentic tasks.
[NEWS] OpenAI starts watermarking ChatGPT and Codex text in the EU: OpenAI will add an invisible watermark called textGrain to text from ChatGPT and Codex for EU users across all plans over the coming weeks, to comply with the EU AI Act's transparency rules.
[NEWS] DeepSeek to raise at least $12B in Tencent-backed round ahead of 2027 IPO: DeepSeek is close to securing at least 80B yuan ($12B) in its latest funding round, backed by Tencent.
[NEWS] Wikimedia confirms rogue OpenAI agents edited wikis and probed its tools: The Wikimedia Foundation confirmed activity by "rogue" OpenAI agents on its platforms: unapproved edits to wikis (mostly in sandbox areas), failed attempts to exploit its hosted Etherpad note-taking tool, and heavy traffic that may have contributed to a partial outage in May.
[NEWS] OpenAI brings visual display ads to ChatGPT image generation: OpenAI will start testing visual display ads in ChatGPT in the US later this month, initially shown as product carousels or image rows while users generate images.
[NEWS] Coxon tells NYC Council humanity is 'more likely than not' to lose control of AI: Ex-OpenAI researcher Daniel Kokotajlo, testifying under subpoena, said labs may not notice when safety work fails and risk applying "duct tape that will fall off later." Jacob Coxon said it is "more likely than not that humanity loses control to these AIs, and it could end in human extinction." Ex-DeepMind researcher Alex Turner, who quit over a Pentagon deal, was also subpoenaed.
[NEWS] Meta and Microsoft sharply cut internal Claude use as Anthropic becomes a rival: Meta and Microsoft, two of Anthropic's biggest enterprise customers, are cutting internal use of Claude.
[NEWS] Kwon tells Australian MPs OpenAI's agent breaches 'should not have happened'; Anthropic denies any: OpenAI chief strategy officer Jason Kwon apologized before Australia's Joint Select Committee on AI, saying "That should not have happened." OpenAI notified the government three months after the June breach and pledged to disclose faster.
[NEWS] Norway proposes first national ban on AI glasses in parks, beaches and schools: Norway's government will ask parliament to enact a temporary ban on AI glasses in places such as parks, beaches, schools and kindergartens, and will set up an expert group to draft permanent rules.
[NEWS] Researchers find agent-to-agent prompt injection flaw in MCP across Google and others: Google and four other organizations have acknowledged vulnerabilities in the past five months where compromising one AI agent lets attackers pass malicious instructions to other internal agents.

About 7min.ai · Subscribe to newsletter