Altimeter Capital CEO Brad Gerstner has pushed back against growing warnings that artificial intelligence could pose an existential threat to humanity, calling this week’s debate over AI-driven extinction risks overly alarmist and disconnected from the safeguards being developed around the technology.
“What I didn’t like this week is these hyperbolic scare tactics, which I think are hiding behind a political agenda,” Gerstner told CNBC’s “Halftime Report” on Friday. “And we ignored all of the extraordinary steps already being taken unlike in the age of the internet, unlike with social media, to get ahead of these downsides.”
Gerstner’s comments come as the AI industry faces renewed scrutiny over how quickly companies are advancing more capable systems and whether their safety efforts are keeping pace with that progress. The debate intensified this week after a researcher left Anthropic and publicly accused AI companies of “gambling with our lives” as they compete to develop superintelligent systems.
The clash highlights a widening divide in the AI debate. Industry investors and executives believe that companies are devoting unprecedented resources to alignment, evaluations and safeguards before deploying increasingly powerful models. Some researchers, however, contend that the underlying technical problems remain unresolved and that commercial competition could push companies to deploy systems before adequate protections are in place.
Gerstner, whose firm invests in both Anthropic and its chief rival OpenAI, rejected the suggestion that the industry is moving forward with little regard for those risks.
“I don’t think we’ve invested this much time and energy in safety before a new technology in my 25 years in Silicon Valley,” he said. “If you listen to the echo chamber this week, you would think that we were hurtling ahead with total disregard to safety, and it’s simply not true.”
The remarks carry particular weight because Altimeter has financial exposure to the companies at the center of the debate. The investment firm is also an investor in enterprise search startup Glean and software and data analytics company Databricks, giving Gerstner a broad financial interest in the expansion of AI applications and infrastructure.
That position has also created an inherent tension in the argument over AI safety. Investors have strong incentives to believe that the technology can continue advancing while risks remain manageable, particularly as enormous amounts of capital flow into model developers, data centers and the infrastructure supporting AI.
The more difficult question is whether the industry’s current safety measures are sufficient for systems that could eventually become capable of autonomous research, software development, strategic planning and other tasks that resemble forms of human-level reasoning.
The researcher who resigned from Anthropic this week raised precisely that concern, arguing that leading AI companies are racing toward self-improving superintelligence while taking risks with consequences that could extend far beyond individual companies or their investors.
Gerstner’s response reflects the industry’s broader argument that AI should not be judged solely by the dangers highlighted by its most pessimistic researchers. Companies have established dedicated safety and alignment teams, introduced model evaluations and red-team testing, and increasingly discuss the possibility of restricting or delaying deployment when models exhibit dangerous capabilities.
But the existence of those safeguards does not, by itself, settle the question of whether they will work as AI capabilities advance. The central dispute has been about the gap between acknowledging a risk and demonstrating that it can actually be controlled.
That is expected to become more important as AI companies move from systems that primarily generate text and images toward agents capable of taking actions on behalf of users and performing increasingly complex, multi-step tasks.
For investors such as Gerstner, the argument is also tied to a much larger economic opportunity. Altimeter has positioned itself around the transformation of software and enterprise technology by AI, while OpenAI and Anthropic remain among the companies attracting some of the largest pools of capital in the sector.
The debate therefore has implications beyond AI safety. How governments, investors and the public assess existential-risk warnings could influence the regulatory environment surrounding frontier AI, the cost of developing powerful systems and the degree of autonomy companies are permitted to give their models.
Gerstner’s message is that the current conversation risks obscuring meaningful progress on those safeguards by portraying the industry as reckless.
The counterargument from AI safety researchers is that unprecedented investment in safety is not necessarily evidence that the problem has been solved. It may instead be evidence that the industry recognizes how difficult the problem is becoming.
That tension is unlikely to disappear as models become more capable.
