# Five AI safety sessions at Disrupt 2026

> TechCrunch Disrupt 2026 features five sessions on AI safety, security, and deployment challenges that founders need to master.
- Title: Five AI safety sessions at Disrupt 2026 · TechCrunch
- Summary: TechCrunch Disrupt 2026 features five sessions on AI safety, security, and deployment challenges that founders need to master. Building AI that people trust…
- Keywords: techcrunch disrupt, ai safety, ai security, enterprise ai, autonomous agents, robotics, physical ai, ai deployment, founders, ai governance, technology, startups
- Source: TechCrunch — https://techcrunch.com/2026/09/22/five-ai-safety-sessions-every-founder-should-have-on-their-techcrunch-disrupt-2026-agenda
- Author: TechCrunch Events
- Published: 2026-09-22T15:00:00+00:00
- Read time: 2 min
- Topics: techcrunch disrupt, ai safety, ai security, enterprise ai, autonomous agents, robotics, physical ai, ai deployment, founders, ai governance, technology, startups
## Disrupt 2026 tackles AI safety
TechCrunch Disrupt 2026 runs October 13-15 in San Francisco with 200+ sessions across six stages. Five sessions focus on AI safety, security, and deployment challenges that founders must solve to earn customer trust.
## From pilots to production
Anthropic Head of Applied AI Cat de Jong will share why some enterprises get real value from Claude while others stay stuck in pilots for 18 months. She works directly with companies moving AI from experiments to critical workflows.
## Agent security is infrastructure
Okta's Ric Smith and NanoCo's Gavriel Cohen will examine how AI agents create new security problems at the infrastructure level. Application-level permissions aren't enough when agents can take autonomous action.
## Enterprise AI needs governance
AWS VP Rudy Mitra, Luta Security CEO Katie Moussouris, and cybersecurity veteran Wendy Nather will discuss security, governance, and observability requirements that enterprises demand before deploying AI in critical systems.
## Physical AI has real stakes
Shield AI's Nathan Michael, GM's Mikell Taylor, and Waabi's Raquel Urtasun will discuss how autonomous systems in vehicles, aircraft, and industrial systems must meet higher safety standards. Mistakes don't stay on a screen.
## Testing can't be an afterthought
When autonomous systems operate in vehicles, aircraft, or critical infrastructure, testing and validation become central to product development. Leaders in autonomy are building companies that earn trust through rigorous safety practices.
## Robots need training data
Nvidia's Les Karpas will explore how robots lack the massive training datasets that accelerated language models and self-driving vehicles. Data pipelines, simulation, and foundation models could help robotics scale and earn trust.
## The ChatGPT moment for robots
Robotics hasn't had its breakthrough moment like large language models did. Solving the data problem could unlock more capable and reliable robots that enterprises will trust to deploy in the real world.
## Trust is the hardest part
Building a smarter model or more capable agent is one challenge. Getting customers to trust it enough to deploy it in their systems may be the hardest part of bringing AI to market.

> You can build a smarter model, a more capable agent, or a robot that does something no one has done before. But getting customers to trust it enough to deploy it may be one of the hardest parts of bringing it to market.
## 10,000 leaders converge
Disrupt 2026 expects 10,000 founders, investors, and tech leaders, plus 250+ speakers and 300+ exhibiting startups. The event runs October 13-15 at Moscone West in San Francisco.
## Save before September 25
Tickets save up to $200 if purchased before September 25 at 11:59 p.m. PT. A second pass is available at 50% off. Founders can also book an exhibit table to demo to 10,000+ tech leaders by October 2.
## Key takeaway

Building AI that people trust to deploy in real systems requires solving safety, security, and reliability challenges alongside technical capability.