Altman, Nadella and Luckey Set Out AI’s Risks at DealBook Summit

Associated Writers

OpenAI’s chief urges human control, Microsoft’s chief expects agents to rework web economics, and Anduril’s co-founder warns on cyberattacks

By Adam Carter

AI leaders at the DealBook Summit on 17 and 18 November 2026 offered predictions on AI safety, the web and cyber defence. OpenAI chief Sam Altman (@sama), Microsoft chief Satya Nadella, and Anduril co-founder Palmer Luckey each took the stage.

Each executive addressed a different risk, and the remarks followed a week of outages and the exit of Nvidia’s founder. TF covered both stories, and the speakers addressed them.

What’s Happening & Why It Matters

Altman and Brockman on Human Control

Altman appeared at the New York Times DealBook Summit on Wednesday, 18 November 2026, per TechCrunch. OpenAI co-founder and President Greg Brockman joined him. OpenAI’s chief said people should stay in charge of AI’s future. He called a gradual handover of decisions to AI “a terrible idea”.

Altman at Summit. (CREDIT: GETTY)

Altman said he and Brockman agree on that goal but differ on how to reach it. He said preventing an accidental or intentional handover requires addressing competitive pressure between companies. TechCrunch links the concern to recursive self-improvement, the idea that AI builds ever more capable successors until people lose control.

Brockman said alignment, safety and security duties grow as models improve. He said chain-of-thought monitoring may lose reliability as capabilities rise. He argued defenders need the most capable models to counter attackers who already have cyber tools.

Altman on Huang and Nvidia

Brockman and Altman spoke about Jensen Huang, who stepped down as Nvidia CEO the week before. Altman said it is hard to imagine the AI industry without Huang. Asked whether Huang could return, Altman declined to give a view and wished Huang and Nvidia well.

Brockman said his family is friends with Huang’s family and that he reached out to support him. TechCrunch reports Nvidia invested $30 billion (about €26.7 billion) in OpenAI earlier in 2026. Nvidia designs the chips that power OpenAI’s technology.

Nearly 200,000 people have signed a petition for Huang’s reinstatement, per TechCrunch. Executives said on Nvidia’s earnings call on 17 November that the company does not plan to reappoint him. TF covered the call in Huang Not Returning.

Nadella on Agents and the Web

Nadella at Summit. (CREDIT: NYT)

Microsoft’s chief predicted that AI agents will reshape internet economics, per TechCrunch. Companies that depend on advertising and human web traffic face the biggest change. Nadella compared the change to the move from portals such as Yahoo to search engines such as Google. He said ad-based companies may need to move toward cost-per-click or transactional models.

He used Amazon as an example, because Amazon blocks AI agents from browsing its catalog and buying for users. Nadella suggested rivals willing to accept agents could gain ground. He named Shopify as one such rival if Amazon keeps its block. TechCrunch notes the remarks are predictions with no tie to an announced Microsoft product.

CNBC’s live blog says Nadella apologised for an extended Outlook outage. CNBC adds that Nadella warned AI makes fast responses to software vulnerabilities critical.

Luckey on Cyberattacks

Anduril co-founder Palmer Luckey (@PalmerLuckey) spoke on Tuesday 17 November, interviewed by Andrew Ross Sorkin, per TechCrunch. TechCrunch says Vice President J.D. Vance cancelled his appearance to attend meetings about the recent tech outages.

Luckey at Summit. (CREDIT: NYT)

Luckey said AI tools let attackers create exploits faster than older systems can defend. He said US adversaries exploit that gap for strategic and financial reasons, with North Korea prominent. He argued AI could favour defenders if defenders move faster than attackers.

TechCrunch says government officials blamed the outages on companies rushing to patch AI-discovered flaws. Luckey focused on organisations running outdated technology. CNBC reports Luckey said no one can control AI after a series of jailbreaks, and that he supports autonomous weapons.

My take: The three speakers share one theme: speed outruns institutions. Altman wants the race managed, Nadella expects web businesses to adapt, and Luckey wants defenders to move faster. Their interests differ, so each prediction deserves a check against that speaker’s business. Nadella and Luckey offered no figures or dates.

TF Summary: What’s Next

The summit gave no new product, policy or date. Altman’s concern about competitive pressure has no fix yet. Nadella’s agent economy depends on retailers such as Amazon changing course.

MY FORECAST: At least one major retailer will open its catalog to third-party AI shopping agents by 30 June 2027. A US government body will publish formal guidance on AI-discovered software vulnerabilities by 31 March 2027.



[gspeech type=full]

Share This Article
Leave a comment