Exciting launch of my second book, AI for Digital Public Infrastructure (DPI) here in #Davos. Extremely grateful to all my mentors, and colleagues with whom i have been involved in AI for 23 years across my three companies - EdCast (Acq. by Cornerstone), PlaySpan(Acq. by VISA) and MobileAria (Acq by WirelessMatrix) as well my work with Government leaders in the Indo-Pacifc. Enjoyed the interview at book launch and i hope you will enjoy watching it, as well as reading the book. a.co/d/da2UPpA #AI #dpi #quad #india #WEF2024

Dario Amodei says AI could wipe out half of entry-level white-collar jobs and push unemployment to 10-20%. Most people heard that and turned it into a jobs debate. But the deeper point is about enterprise risk. If AI is already moving into the center of what junior analysts, consultants, lawyers, finance teams and operators do, then companies are not just adopting software anymore. They are delegating judgment. That changes the whole operating model: - who reviews the output - what gets logged - what gets approved - what gets blocked - what happens when the model is wrong The companies that win will not be the ones that deploy AI fastest. They will be the ones that can prove their AI systems are safe enough to trust. Full article below. If you're building with agents, don't just read it. Star/fork TrustModel and use it as an open-source eval governance layer before giving AI real permissions: github.com/karlmehta/trustmo…

Dario Amodei was asked if businesses, policymakers and governments are doing enough to prepare for AI. His answer: "No." This is one of the clearest warnings from the person building one of the most important AI labs in the world. What he is saying is simple: - AI capability is moving on a stable exponential curve - public opinion keeps swinging between panic and hype - companies are deploying before they have real assurance systems - policymakers are reacting after the fact - most teams still treat evals like a compliance checkbox If you are putting AI into production, the question is not "can it answer the prompt?" The question is whether you can test it, monitor it and govern it before it creates risk at scale. Full article below. If you're building with agents, don't just read it. Star/fork TrustModel and use it as an open-source eval governance layer before giving AI real permissions: github.com/karlmehta/trustmo…

Anthropic does not wait for Claude to fail in the wild. They try to break it first. Adversarial prompts. Misuse scenarios. Autonomy stress tests. Deception probes. Tool-use failures. Most companies shipping AI agents do the opposite. They run one impressive demo. Then give the agent access to customers, files, tickets, CRM, code, or APIs. That is the entire market gap. The frontier labs have red teams. Most builders have vibes. TrustModel turns that lesson into infrastructure developers can actually use: Eval before release. Monitor drift in production. Govern outputs against policy packs mapped to EU AI Act, NIST, ISO, NYC LL144, and OWASP. Check out the repo: github.com/karlmehta/trustmo…