Anthropic is hiring someone to investigate its own employees

Anthropic is advertising for an Insider Risk Investigator at $245,000–$305,000, based in Boston, NYC or DC. The job: triage detection alerts, run end-to-end insider investigations, comb through SIEM, DLP and UEBA logs, monitor external threats targeting employees — and conduct "sensitive interviews of employees or other involved parties." It wants 5–8 years in corporate investigations and 3–5 years of investigative interviewing. Counterintelligence experience and familiarity with nation-state TTPs are explicitly called out, as is government-setting interview work. It's not a one-off: Anthropic has several insider-risk openings live, including forensics and detection-platform variants.

The timing is doing a lot of work here. On 3 August, Axios reported — via an unnamed source — that Dario Amodei privately worries new hires are showing up for the paycheck rather than the mission. Anthropic reportedly leads frontier labs on retention at 80%, but it has scaled from seven people to 2,500+, which is well past the point where founder networks vouch for everyone walking in the door.

Read charitably, this is what any lab holding frontier model weights and trade secrets eventually builds, and rivals run similar programs more quietly. Read less charitably, a company whose pitch is trustworthy AI is now staffing up to decide which of its own people it can trust.

Why it matters: the AI talent war has escalated from poaching to counterintelligence — model weights are now treated like state secrets, and employees like an attack surface.