Diller Trusts Altman But Says Trust Won't Matter as AGI Nears

Barry Diller, the billionaire chairman of IAC and Expedia Group, has publicly defended OpenAI CEO Sam Altman as a trustworthy leader — while simultaneously arguing that personal trust is fundamentally inadequate as a safeguard against the risks of artificial general intelligence (AGI). The paradox at the heart of Diller's message captures a growing tension across the tech industry: even the most well-intentioned leaders cannot serve as the sole guardrail against a technology that could reshape civilization.

Diller's comments arrive at a pivotal moment for OpenAI, which is navigating a controversial transition from a nonprofit to a for-profit structure, escalating competition with rivals like Google DeepMind, Anthropic, and Meta AI, and intensifying scrutiny over its safety practices.

Key Takeaways

• Barry Diller expressed personal trust in Sam Altman's leadership at OpenAI
• Despite that trust, Diller warned that 'trust is irrelevant' as AGI development accelerates
• He called for structural guardrails that go beyond any single leader's judgment
• The comments reflect broader industry anxiety about AGI timelines and governance
• OpenAI's ongoing restructuring adds urgency to the governance question
• Diller's stance aligns with a growing chorus of voices demanding systemic AI oversight

A Media Mogul Weighs In on AI's Biggest Question

Barry Diller is no stranger to disruptive technology. The 83-year-old media executive built his career navigating transformative shifts in entertainment, from broadcast television to the internet age. His business empire, which includes the conglomerate IAC and its portfolio of digital brands, gives him a vantage point few possess when evaluating how technology reshapes industries.

Diller's defense of Altman is notable because it comes from outside the traditional Silicon Valley echo chamber. He is not an AI researcher or a venture capitalist with a financial stake in OpenAI's success. His endorsement of Altman's character carries the weight of an independent observer who has watched countless tech leaders rise and fall.

Yet the crux of Diller's argument is that character alone cannot govern a technology as powerful as AGI. 'Trust is irrelevant,' he stated plainly, drawing a sharp line between personal confidence in a leader and the systemic risks posed by superintelligent AI systems. This distinction matters enormously as the AI industry barrels toward increasingly capable models.

Why Personal Trust Falls Short in the AGI Era

The concept Diller raises cuts to the heart of a fundamental governance problem. Throughout history, societies have learned — often painfully — that relying on the good intentions of individuals is no substitute for institutional safeguards. Nuclear weapons, financial markets, and pharmaceutical development all required regulatory frameworks precisely because the stakes were too high to leave to personal judgment.

AGI — broadly defined as AI that can match or exceed human cognitive abilities across virtually any domain — represents a similar inflection point. Leading AI labs including OpenAI, Google DeepMind, and Anthropic have all acknowledged that AGI could arrive within this decade, with some estimates placing it as early as 2027 or 2028.

Consider the governance challenge this creates:

• A single CEO's values could shift over time, or they could be replaced
• Corporate incentives may conflict with public safety as commercial pressure mounts
• The speed of AI development may outpace any individual's ability to foresee consequences
• Competitive dynamics between labs create pressure to cut corners on safety
• No single person — however trustworthy — can fully comprehend the implications of AGI

Diller's framing implicitly acknowledges all of these realities. He is not questioning Altman's motives; he is questioning whether motives are sufficient.

OpenAI's Governance Crisis Adds Context

Diller's remarks gain additional weight when viewed against OpenAI's turbulent recent history. In November 2023, OpenAI's board famously fired Altman before reinstating him just days later — a saga that exposed deep fractures between the company's safety-focused mission and its commercial ambitions.

Since then, OpenAI has undergone significant leadership changes. Several key safety researchers, including co-founder Ilya Sutskever and safety lead Jan Leike, departed the company in 2024, raising questions about whether safety considerations were being sidelined in favor of rapid product development.

OpenAI is also in the process of converting from its original capped-profit nonprofit structure to a full for-profit corporation, a move that has drawn criticism from former board members and AI safety advocates. The restructuring could be worth upwards of $300 billion based on recent valuation discussions, creating enormous financial incentives that may not always align with cautious AGI development.

These developments underscore Diller's point: even if Altman is personally trustworthy today, the organizational and market forces surrounding him are constantly shifting. Structural guardrails — not personal assurances — are what endure.

The Broader Industry Echoes Diller's Concerns

Diller is far from alone in calling for systemic AI governance. A growing number of prominent voices across the tech industry and policy world have made similar arguments, though often from different angles.

Anthropic CEO Dario Amodei has repeatedly argued for what he calls 'responsible scaling policies' — formalized frameworks that tie safety testing to capability thresholds. Unlike personal trust, these policies are designed to function regardless of who leads the company.

Yoshua Bengio, the Turing Award-winning AI researcher, has called for international governance mechanisms modeled on nuclear nonproliferation treaties. The EU AI Act, which took effect in 2024, represents the most comprehensive regulatory framework to date, though critics argue it still lacks teeth when it comes to frontier AGI systems.

Meanwhile, the U.S. government has taken a more fragmented approach. Executive orders from the Biden administration established some reporting requirements for frontier AI models, but comprehensive legislation remains elusive. The current political environment suggests that any sweeping AI regulation faces significant headwinds in Congress.

Key governance proposals currently under discussion include:

• Mandatory safety evaluations before deploying frontier AI models
• Independent auditing requirements for AGI-capable systems
• International coordination frameworks similar to nuclear oversight bodies
• Liability frameworks that hold AI developers accountable for harms
• Compute governance — controlling access to the massive hardware needed for AGI training
• Whistleblower protections for AI safety researchers at major labs

What This Means for the AI Industry

Diller's comments, coming from a respected business leader outside the AI bubble, signal that concerns about AGI governance are going mainstream. This has practical implications for multiple stakeholders.

For AI companies, the message is clear: self-regulation and appeals to leadership character will not satisfy public expectations indefinitely. Companies that proactively embrace transparent governance structures may gain a competitive advantage as regulatory pressure mounts. OpenAI, Anthropic, Google DeepMind, and others will likely face increasing demands to demonstrate structural — not just rhetorical — commitments to safety.

For investors, the governance question introduces a new dimension of risk. A company's AGI safety practices could become a material factor in valuations, particularly as regulatory frameworks take shape. The potential for liability exposure from AGI-related harms is a wildcard that no current financial model adequately captures.

For policymakers, Diller's framing offers a useful rhetorical tool. The argument that 'trust is irrelevant' sidesteps the politically charged question of whether specific tech leaders are good or bad actors, focusing instead on the structural need for oversight regardless of who is in charge.

For developers and engineers working at AI labs, the governance debate directly affects workplace culture, safety priorities, and the ethical weight of day-to-day decisions. The departures from OpenAI's safety team suggest that this tension is already shaping career decisions across the industry.

Looking Ahead: The Race Between Capability and Governance

The fundamental challenge Diller identifies — that trust in individuals cannot scale to match the risks of transformative technology — will only intensify as AI capabilities advance. OpenAI's GPT-5, expected later in 2025, is rumored to represent a significant leap in reasoning capabilities. Google DeepMind's Gemini models continue to push boundaries. And dozens of well-funded startups are racing to build increasingly powerful systems.

The critical question is whether governance frameworks can keep pace with this acceleration. History offers mixed precedents. Nuclear governance emerged only after catastrophic use. Financial regulation typically follows crises. The AI community has a narrow window to break this pattern and establish meaningful guardrails before they are desperately needed.

Diller's contribution to this conversation is deceptively simple but profoundly important. By separating the question of personal trust from the question of structural safety, he clarifies the debate in a way that should resonate across political and ideological lines. Sam Altman may indeed be trustworthy. But building the future of civilization on any single person's trustworthiness is, as Diller suggests, a bet no rational society should make.

The coming 12 to 24 months will be decisive. As AGI timelines compress and commercial stakes escalate, the gap between what trust can provide and what governance must deliver will only widen. Whether the industry — and the world — rises to meet that challenge remains the defining question of the AI era.