Skip to main content
World Today News
  • Home
  • News
  • World
  • Sport
  • Entertainment
  • Business
  • Health
  • Technology
Menu
  • Home
  • News
  • World
  • Sport
  • Entertainment
  • Business
  • Health
  • Technology
OpenAI's Altman and Brockman Say Humans Must Control AI Agents to Unlock Enterprise Adoption

OpenAI’s Altman and Brockman Say Humans Must Control AI Agents to Unlock Enterprise Adoption

November 19, 2026 Business Desk Business

OpenAI co-founders Sam Altman and Greg Brockman told business leaders at the DealBook Summit in New York City this week that autonomous AI agents must remain under human control, addressing how enterprises can deploy the systems without exposing themselves to rogue behavior.

The interview, hosted by New York Times columnist Andrew Ross Sorkin, comes as security has replaced capability as the top concern holding back corporate AI adoption.

The New York Times recorded the conversation live at its annual DealBook Summit, which gathers business and policy leaders for onstage interviews. A video of the exchange, titled “Why Sam Altman and Greg Brockman Say Humans Must Stay in Control of A.I.,” runs about one minute and fifteen seconds and is available through the Times’ DealBook coverage.

The stakes extend well beyond one company’s stage appearance. As enterprises wire large language models into decision-making workflows, the risk of an agent acting outside its intended instructions — even unintentionally — has become a boardroom-level issue, and the two co-founders framed safety as the condition on which the entire enterprise market depends.

Why Sam Altman and Greg Brockman Say Humans Must Stay in Control of AI — The New York Times

Altman said OpenAI is prioritizing interpretability tools to make sure system actions align with human intent, according to a report on the appearance from BNewsO. He said that current models require significant oversight while future iterations will operate with greater autonomy — a shift, he argued, that forces a redesign of traditional IT security protocols.

Security, not capability, is the adoption bottleneck

The competitive backdrop explains the timing. Rivals including Anthropic and Google DeepMind have announced their own enterprise-focused security suites, and OpenAI is positioning itself around granular control over agent permissions, BNewsO reported.

The numbers show why executives are listening. Roughly 67% of enterprise leaders cite security as their top barrier to full-scale AI adoption, according to industry data cited by BNewsO, and industry estimates put the market for AI security solutions above $2 billion by 2026.

Brockman said OpenAI’s infrastructure is being hardened to withstand adversarial attacks targeting the underlying model weights, and he framed safety as a continuous process of evaluation and refinement rather than a binary state, per BNewsO’s account of the discussion.

Both executives acknowledged that no system is entirely risk-free. Their argument, as BNewsO reported it, is that the economic benefits of deployment must be weighed against potential failures, and that transparent incident reporting is essential to keeping corporate partners on board.

Three ways the agent-security debate reshapes enterprise AI

  • Compliance becomes a product feature. BNewsO reports OpenAI expects to release new compliance tools designed specifically for large-scale enterprise contracts in the coming months, turning safety controls into a selling point rather than a constraint.
  • IT security protocols get rebuilt around agents. With future models expected to act with greater autonomy, the traditional perimeter-based security model gives way to real-time monitoring of agent behavior through interpretability layers, per BNewsO’s reporting on Altman’s remarks.
  • Regulators move in parallel. The executives’ push for transparency aligns with regulatory trends in the United States and Europe, where lawmakers are drafting guidelines for high-risk AI applications, BNewsO reported.

Permission structures meet procurement reality

Enterprise clients are already demanding stricter permission structures to prevent unauthorized data access by AI tools, according to BNewsO.

The shift from raw capability to reliable, secure integration marks a change in how the industry sells itself. OpenAI’s strategy, as BNewsO characterized it, relies on demonstrating that advanced AI can be both powerful and predictable, with leadership committed to working alongside governments and private-sector stakeholders on industry-wide standards.

The two accounts of the appearance differ in emphasis. The New York Times framed the session around the co-founders’ argument that humans must stay in control of AI, presenting it as a conversation about balancing safety and risk; BNewsO leaned harder into the commercial framing, tieing the remarks to enterprise security suites, market estimates and compliance tooling. Neither account publishes a full transcript.

For chief information officers and security teams, the practical takeaway is timing: OpenAI’s compliance tooling has not shipped yet, and the executives themselves described safety as unfinished work. BNewsO reported that stakeholders are watching closely to see whether these commitments translate into measurable security improvements.

OpenAI’s path to this stage began in December 2015, when it was founded as a nonprofit with Elon Musk and Sam Altman as co-chairs; Musk resigned in 2018, and a for-profit subsidiary followed in 2019. The company’s release of ChatGPT in November 2022 accelerated a global AI boom, making generative AI a mainstream corporate concern.

Its safety record has drawn scrutiny along the way. In November 2023, OpenAI’s board removed Altman as CEO, citing a lack of confidence, before reinstating him five days later after a board reconstruction. In 2024, roughly half of the company’s AI safety researchers left, often citing a deprioritization of safety — a history that gives the co-founders’ latest assurances particular weight.

The full interview is available through the Times’ DealBook coverage, which continues to publish conversations from the summit with the leaders at the center of the AI governance debate.

More on this story: Satya Nadella Envisions a Future With 20 Million AI Agents at Microsoft · Anthropic, the Maker of the Claude Chatbot, Has Priced Its Initial Public Offering at $83 per Share

More on AI Agents: Consumer Reports Suggests Engineering Loyalty Into AI Agents · Estonia Plans Digital Sub-Identities for AI Agents to Use Services

Share this:

  • Share on Facebook (Opens in new window) Facebook
  • Share on X (Opens in new window) X

Keep reading

  • HMD Teases PureView Return With Schneider-Kreuznach Optics, Possible Periscope Lens
  • Qonto Obtains Crédit Institution Licence, Plans Direct Lending and Savings

Related

Search:

World Today News

World Today News is your trusted source for global journalism — breaking headlines, in-depth analysis, and reporting from around the world.

Quick Links

  • Privacy Policy
  • About Us
  • Accessibility statement
  • California Privacy Notice (CCPA/CPRA)
  • Contact
  • Cookie Policy
  • Disclaimer
  • DMCA Policy
  • Do not sell my info
  • EDITORIAL TEAM
  • Terms & Conditions

Browse by Location

  • GB
  • NZ
  • US

Connect With Us

© 2026 World Today News. All rights reserved. Your trusted global news source directory.
For contact, advertising, copyright, issues email: [email protected]

Privacy Policy Terms of Service