Dell: 4% Test AI Agent Incidents

Key Takeaways

• Dell reports only 4 of 25 survey respondents tested an AI agent incident scenario
• Dell uses identity controls, logging and traceability to manage AI agent risk
• Guardrails set upfront can enable companies to take more risk, not less

Dell says AI agent guardrails enable faster and safer business operations

Dell says companies should build security and resilience into AI agents from the start. Only 4 out of 25 survey respondents had tested an AI agent incident scenario. Dell uses tight identity controls, logging and traceability for its agents. Guardrails set upfront can let companies take more risk, not less.

Workiva event highlights need to tie AI controls to workflow design

Enterprise AI governance is moving into data, workflows and controls that decide what automated systems can do. Companies need AI results that are traceable to sources and backed by clear accountability. Governance and workflow redesign must advance together so speed becomes an asset, not a risk. AI readiness must be evaluated process by process.

How to spot AI-generated videos before you share them

A Fort Bend County judge's campaign video was found to use AI-generated images of Black voters and was later removed. Experts say AI content is everywhere on social media and getting harder to detect. Look for visual inconsistencies, disappearing objects, text errors, stiff motion and robotic tones. Even with those signs, AI is improving fast, so spotting fakes is not always simple.

Industrial Intelligence could be the next big growth driver in manufacturing

Physical AI lets machines perceive conditions, reason about them and take action on the factory floor. Agents can spot patterns, predict problems and decide the best course of action. Simulation lets manufacturers test changes in a digital environment before touching real equipment. The goal is to help people become more capable by giving machines better ways to act on knowledge.

AI models still lag behind existing scores for COPD prognosis

AI models have not yet outperformed existing clinical scores for predicting COPD outcomes. Machine learning models for predicting flare-ups and long-term prognosis are not ready for routine clinical decisions, said Jean Bourbeau, MD. Home monitoring may still help track disease activity. A new COPD treatment framework focuses on disease activity and stability.

Epic Games opposes AI reverse engineering despite pushing AI

Epic Games, the maker of Fortnite, is criticizing AI reverse engineering even though it has pushed AI use in its own products. CEO Tim Sweeney pushed back on reports about this stance. Microsoft has also targeted AI reverse engineering while promoting AI use internally. Both companies face criticism for hypocrisy in their AI policies.

AI transforms UAE car insurance with personalized pricing

AI is changing car insurance in the UAE by analyzing up to 100 factors per policy. The technology replaces standard pricing with personalized premiums based on driver behavior, location, and weather. Insurers now offer same-day settlements and products tailored to how people actually drive. UAE leads in adopting these digital technologies.

Pegasus founder urges governments to build sovereign AI

Shalev Hulio, co-founder of NSO Group which created Pegasus spyware, advises governments to develop their own AI models. He says nations should not rely on US AI companies for sensitive data. Hulio's company Dream has raised $400 million and offers AI tools for government cybersecurity. The company has sold products to over a dozen countries.

University of Michigan hosts AI healthcare policy panel

University of Michigan and European professors discussed AI in healthcare at a Ford School of Public Policy event. About 40 attendees heard panelists address regulation and human-in-the-loop systems. Law professor Nicholson Price said training AI to be unbiased is feasible with proper data. The panel explored how AI can personalize healthcare while addressing concerns about government oversight.

Blind veterans receive AI smart glasses at UT Austin

Blind and visually impaired veterans received Ray-Ban Meta AI glasses at the University of Texas at Austin's Dell Campus. Meta pledged 130,000 pairs of glasses to blind veterans across the U.S. The glasses use a wearable camera and audio with AI to help users navigate surroundings hands-free. Veterans received individual training to set up and use the devices.

Philosopher explores friendship with AI

Valerie Tiberius, a philosophy professor at the University of Minnesota, discussed how people are forming friendships with artificial intelligence. The talk examined how AI apps like ChatGPT, Gemini, and Claude are changing the way people connect. The discussion took place as new data showed the growing number of AI-based apps available. The event was reported on October 27, 2026.

Sources

• Dell sees AI agent guardrails as a path to greater autonomy
• Enterprise AI governance ties controls to workflow design
• Don’t get fooled, here’s how to spot an AI-generated video before you share it
• Industrial Intelligence: The next growth opportunity for manufacturing
• AI Not Yet Ready for COPD Prognosis, With Jean Bourbeau, MD
• Epic Games Is Suddenly Anti-AI Now That It's Being Hurt by It
• How AI is changing UAE car insurance: Faster claims, fairer pricing and products built around how you actually drive
• ‘You want to control your own AI’: Pegasus spyware founder tells governments to push for sovereign AI and national security over reliance on US AI
• UMich and European professors discuss AI in healthcare
• Blind veterans get Ray-Ban Meta AI glasses, training at UT Austin Dell Campus
• Local philosopher examines definition of friendship by looking at relationships with AI