AI News Roundup: Anthropic's IPO Push, Google Buys a Data Center Site, and AI Agent Security

Here is the AI news from the last day or so that matters for small businesses using AI. Every item links to its source.

Anthropic keeps its IPO on track. Investors cited by the Financial Times say Anthropic is aiming for a $1.25 to $1.75 trillion valuation and hopes to raise $80 to $100 billion before Thanksgiving. (Kill The AI)

Google buys Crusoe's Wyoming data center site. Google took over the unfinished campus and plans to invest $17 billion there, a sign of how hard huge AI builds are for smaller developers. (Kill The AI)

AI agent security gets a price tag. Palo Alto Networks is in talks to buy Clutch Security, which protects the passwords and tokens AI agents use, for $500 million in cash. Separately, security firm Apiiro reported that Claude Code's "Don't Ask Again" option can store secrets in plain text, with leaked keys found in public GitHub repos. (Kill The AI, LLMnet briefing)

Voice keeps spreading. Google Antigravity added a voice mode built on Gemini Live, so developers can plan work and approve actions by speaking, in several languages, and interrupt it mid-reply. (LLMnet briefing)

Cheaper AI video and a new federal task force. Gemini Omni 1.2 Flash is generally available with 15-second clips and a 30% price cut, and the Justice Department set up a task force to prosecute misuse of advanced AI. (BYOBot)

What it means for you: voice is becoming a normal way to work with software, and the tools your AI uses need the same care as an employee's keys. If you want an AI voice agent that answers every call and keeps your data safe, see Ozkra or contact us.