Anthropic’s new consulting venture: Why are AI model companies moving into the services business? | Explained

Yesterday, Anthropic announced progress on a system that cryptographically proves which model generated a specific output — and confirms the prompt wasn't tampered with after the fact.

I've spent the last eighteen months watching boards and risk committees ask the same question: "How do we know this AI output is authentic?" At Abilitix Consulting, we've seen financial services clients build entire parallel review processes because they can't trust the chain of custody between prompt, model, and output. One bank we worked with last quarter employed seven people just to manually verify that customer-facing AI responses hadn't been altered between generation and delivery.

That verification gap is expensive, but the lesson runs deeper: provenance isn't about the technology being trustworthy — it's about being able to demonstrate trust to a regulator, an auditor, or a customer who's been wronged.

Anthropic's approach (they call it "provable inference") creates a cryptographic signature that binds three elements together: the exact model version, the precise prompt, and the generated output. Change any one element, and the signature breaks. This isn't theoretical. The same technique already secures software supply chains and ballot systems.

The application for enterprises is immediate. APRA CPG 234 requires institutions to trace material risk decisions back to their inputs. ISO/IEC 42001 demands documented accountability for automated outputs. Right now, that traceability lives in log files and screenshots — evidence a first-year intern could alter with a text editor.

A cryptographically signed audit trail removes "trust me" from the equation.

Do your current AI vendors give you tamper-proof evidence of what the model actually said — or just a database entry you're supposed to believe?

#AIGovernance #APRACPG234 #EnterpriseAI #TrustedAI #AIAuditability

📩 The full weekly breakdown lands in AI Governance and Compliance — Daily Brief → https://amplyfy.app/wire/subscribe/200