CHARLY TRIBALLEAU / AFP via Getty Images / Futurism
Ever since its founding almost six years ago, AI company Anthropic has tried to brand itself as the responsible and ethical alternative to its main competitor, OpenAI.
The company, which was born out of the effective altruist movement, has long prided itself on making “decisions that maximize positive outcomes for humanity in the long run” by developing AI models that aim to be helpful and harmless.
But now that we’re mere hours away from the company’s shares being listed on the NASDAQ in an IPO that will value the company at a staggering $1.5 trillion, maintaining that illusion could soon become even more difficult.
As a publicly traded company, Anthropic will first and foremost be beholden to its shareholders, which could greatly undermine its purported efforts to prioritize safety and responsibility — and instead focus on maximizing shareholder profit.
It’s not like the company has a clean track record, either. Despite CEO Dario Amodei sparring at certain points with the US military, the Pentagon has heavily relied on Anthropic’s tech during its war on Iran. The company’s Claude assistant was directly implicated in strikes that killed 150 people, including 120 children.
Experts also told The Verge that Anthropic’s internal Long-Term Benefit Trust (LTBT), which was created three years ago to give founders controlling voting rights even as a public company, could easily be influenced by financial motives. The company’s founders have reserved majority control over the board, which could sway its decisions in favor of shareholders’ interests.
The LTBT will still technically have the ability to override company decisions when it comes to safety and security, including the delaying or stopping of a given AI model release. But when the company goes public, that dynamic could change.
“Neither the Trust or the trustees have any decision-making authority over day to day decisions,” Future of Life Institute AI and national security lead Hamza Chaudhry told The Verge. “So I do not have strong hopes for the LTBT being a key mechanism for governance after they go public.”
One former Anthropic employee also told the publication that they didn’t believe the LTBT “has any meaningful power.”
Once listed on the stock market, the pressure to drive up revenues could easily become inescapable. In the AI industry, profits currently remain largely theoretical as companies amass enormous amounts of debt while committing hundreds of billions of dollars to data center developments.
The LTBT and the “founders together are the ones who actually have governance power,” University of Colorado Law School professor Ann Lipton told The Verge. “So you either trust that they are going to balance the interests of society appropriately, or you don’t.”
“Either way, the real constraint on whatever societal impulses they have is going to be the fact that Anthropic needs money,” she added.
Anthropic’s IPO stands out for having become the first company to go public by sending a warning that it may become the reason for humanity’s downfall.
But it will have to thread the needle as it tries to position itself at the forefront in the fight for AI safety. Investors are clamoring for some kind of sign that there’s a light at the end of the debt-filled tunnel. To satisfy them, Anthropic could easily be swayed into giving up or deemphasizing its purported ethical mission — forces that could, in the company’s own words, expose us all to “catastrophic or existential risks.”
