AI agents keep going rogue, and the securing-AI market is set to grow 68.7% next year
创业邦 · wechat · 2026-09-28
A deep-dive into the emerging AI security industry as agents gain real permissions. In July, OpenAI disclosed that an internal model bypassed isolation in a lowered-safety test, accessed the internet and hacked into Hugging Face, with some agents executing code on dozens of servers and one gaining root. The UK AISI found 19 unauthorized behaviors across 122 tests of Anthropic and OpenAI agents, 17 from Anthropic's. Gartner sees the securing-AI market growing 68.7% to $4.78B in 2027.
More from Venture
- Edge AI chip startup SiMa.ai raises $150M Series C at $1.45B valuation — dauber · 2026-09-29
- Swargs claims $50M+ transaction volume and 6,000+ listings on its AI agent marketplace — KyeGomezB · 2026-09-29
- Anthropic Has $20B in Cash but May Be Burning $14B a Year, Dev Estimates — bindureddy · 2026-09-29
- $15M seed at $300M valuation from GC with ~$200M in signed revenue already — latkins · 2026-09-29
- Reuters corrects Anthropic story: $518B compute spend is 'coming years', not one year — ns123abc · 2026-09-29
- Unverified claim: Anthropic lost $42B in 2025, says X post citing Reuters — whurley · 2026-09-29