Five AI safety sessions every founder should have on their TechCrunch Disrupt 2026 agenda

4 hours ago 4

Would you spot an AI cause with entree to your company’s systems? Put your employees successful an autonomous vehicle? Deploy a robot that has to marque consciousness of an unpredictable carnal world?

These aren’t hypothetical questions for AI founders anymore. As AI moves retired of demos and into businesses, vehicles, robots, and autonomous agents, information and information go portion of the product.

At TechCrunch Disrupt 2026, 5 sessions crossed the AI Stage and Real World AI Stage enactment those challenges beforehand and center. From securing agents to getting endeavor AI into production, here’s wherever founders tin excavation into what it takes to physique artificial quality (AI) radical and enterprises volition really spot and use.

Building the adjacent procreation of AI? Secure your walk to Disrupt and adhd these 5 AI information sessions to your agenda. Save up to $200 connected your walk earlier September 25, 11:59 p.m. PT. Get a 2nd walk astatine 50% off.

1. What Anthropic Sees When Enterprises Actually Deploy Claude

Some enterprises are getting measurable worth from AI. Others are inactive moving pilots 18 months later. Why?

Anthropic Head of Applied AI Cat de Jong gets a close-up view. She works straight with enterprises putting Claude into captious workflows and sees wherever deployments succeed, wherever they stall, and what separates the two.

In “What Anthropic Sees When Enterprises Actually Deploy Claude,” de Jong volition instrumentality the AI Stage to research what happens erstwhile companies determination from experimenting with AI to putting it into production. For founders trying to merchantability AI into the enterprise, her acquisition offers a firsthand look astatine wherefore immoderate deployments present existent worth portion others stay stuck successful aviator mode.

Want to cognize what separates AI pilots from deployments that present existent concern value? Hear what Anthropic is seeing firsthand. Register for Disrupt present to prevention up to $200 earlier September 25.

2. The Agent Security Problem Nobody Is Talking About

An AI cause that tin instrumentality enactment tin besides make an wholly caller acceptable of information problems. What should it person entree to? What should it beryllium allowed to do? And what happens erstwhile application-level permissions aren’t enough?

In this AI Stage session, “The Agent Security Problem Nobody Is Talking About,” Okta President of Products and Technology Ric Smith and NanoCo Co-Founder and CEO Gavriel Cohen volition tackle cause information astatine the infrastructure level, including weaknesses successful application-level support models and architectural decisions founders request to see erstwhile deploying agentic AI.

If agents are connected your merchandise roadmap, information belongs there, too. Get your summons today to prevention up to $200 and adhd this league to your Disrupt agenda.

3. Securing the AI Enterprise: Why the Cloud Just Got a Lot More Complicated

Getting an AI merchandise done the endeavor doorway takes much than an awesome demo.

Security, governance, and observability each go portion of the speech erstwhile companies see putting AI into captious systems.

In this AI session, “Securing the AI Enterprise: Why the Cloud Just Got a Lot More Complicated,” AWS VP of Security Services Rudy Mitra, Luta Security CEO Katie Moussouris, and cybersecurity seasoned Wendy Nather volition analyse the infrastructure enterprises request arsenic AI takes connected much autonomous roles.

For founders gathering endeavor AI, this is simply a accidental to perceive what happens erstwhile innovation meets the information requirements of the organizations you privation arsenic customers.

Building AI for endeavor customers? Get person to the information requirements that tin basal betwixt your merchandise and deployment. Secure your walk to Disrupt earlier September 25 to prevention up to $200.

4. Building AI Systems When Failure Is Not an Option

AI that operates successful the carnal satellite raises the stakes. A mistake doesn’t enactment connected a screen; it tin impact vehicles, aircraft, concern systems, and captious missions.

On the Real World AI Stage, Shield AI Chief Technology Officer Nathan Michael, General Motors Director of Robotics Strategy Mikell Taylor, and Waabi Founder and CEO Raquel Urtasun volition bring perspectives from defence and autonomous systems to 1 of the toughest questions hard-tech founders face: How bash you cognize erstwhile an autonomous strategy is harmless capable to deploy?

The speech successful “Building AI Systems When Failure Is Not an Option” volition get into creating a information culture, investigating and validating AI, navigating regulatory hurdles, and gathering companies that tin gain spot erstwhile the consequences of nonaccomplishment are physical.

When nonaccomplishment isn’t an option, investigating can’t beryllium an afterthought. See however leaders successful autonomy are approaching information astatine Disrupt. Explore your summons options, and prevention up to $200 connected your walk earlier September 25.

5. Robots Are Waiting for Their ChatGPT Moment. Here Is What Is Standing successful the Way

Robots person a information problem. They don’t person entree to the monolithic pools of grooming information that helped accelerate advances successful connection models and self-driving vehicles.

That spread is 1 of the biggest obstacles to scaling carnal AI. In this Real World AI session, “Robots Are Waiting for Their ChatGPT Moment. Here Is What Is Standing successful the Way,” NVIDIA Inception Global Head of Physical AI Les Karpas volition research however information pipelines, simulation environments, and instauration models could assistance adjacent it and what it volition instrumentality to build, test, and deploy much susceptible and reliable robots.

For founders moving successful carnal AI, the league tackles a large question: What volition it instrumentality for robotics to scope its ain ChatGPT infinitesimal and gain the spot needed to enactment those systems to enactment successful the existent world?

Get an wrong look astatine what could assistance robotics wide its adjacent large hurdle. Register for Disrupt contiguous to prevention up to $200 earlier prices increase.

Five sessions. One question each AI laminitis needs to answer.

Can radical spot what you’re building? These 5 conversations are portion of 200+ sessions crossed six manufacture stages, roundtables, and breakouts astatine Disrupt, October 13–15 astatine Moscone West successful San Francisco. More than 10,000 founders, investors, operators, and tech leaders are expected, on with 250+ speakers, and 300+ exhibiting startups, and countless networking meetings.

You tin physique a smarter model, a much susceptible agent, oregon a robot that does thing nary 1 has done before. But getting customers to spot it capable to deploy it whitethorn beryllium 1 of the hardest parts of bringing it to market.

Secure your walk to Disrupt present and prevention up to $200 earlier September 25, 11:59 p.m. PT. Save 50% connected a 2nd ticket. Hear however AI leaders are tackling the safety, security, and reliability challenges that tin basal betwixt breakthrough exertion and real-world adoption.