Five AI safety sessions every founder should have on their TechCrunch Disrupt 2026 agenda
THE SO WHAT
AI safety getting its own track with Anthropic, NVIDIA, AWS, and Waabi on stage shows governance is moving from research niche to mainstream founder concern. If you’re raising or selling into enterprise, arrive with a concrete safety story—evals, red-teaming, and incident response, not just a model card.
READ THE SOURCE
MORE FROM THE WIRE
Applied AICould Meta’s viral Muse app be the company’s ChatGPT moment?
If Muse is outpacing ChatGPT’s early download curve, the assistant surface is consolidating into a few consumer gateways where distribution and default status matter more than raw model quality. For consumer-facing products, you now have to assume Meta will be one of the primary AI entry points you integrate with or compete against.
Applied AIParloa says it is partnering with Zalando on AI customer conversations
If Zalando is rolling Parloa’s AI across European customer conversations under the new AI Act, disclosure and labeling move from “nice to have” to regulated requirement. Any team deploying AI in customer support in the EU now has to treat transparency and auditability as core product features, not compliance afterthoughts.
Applied AISalesforce Ethics Officer on AI Integration, New Book
Salesforce is framing AI upside as a function of how you design human–machine collaboration, not just model strength. That’s a cue for operators to invest in workflow redesign and change management as first-class line items in AI programs, not soft add-ons.
Applied AII tested ChatGPT against Copilot in Microsoft Word
If Copilot excels when you know exactly what you want and ChatGPT helps more when you don’t, assistant choice becomes a function of task ambiguity. Teams should stop treating “AI in Office” as a monolith and instead map tools to workflows—spec writing vs. ideation vs. editing—accordingly.