Anthropic is facing growing criticism over the safety, environmental and community costs of its expansion, even as the artificial intelligence company prepares for a potential initial public offering. The concerns range from warnings that advanced AI could threaten humanity to disputes over data center construction, energy use, pollution and the concentration of corporate voting power.
Safety concerns inside Anthropic
Some Anthropic employees have taken the company’s safety concerns to an extreme. Researcher Jacob Coxon resigned from Anthropic in September, saying he believed that without a greater focus on bringing AI models under human control, humanity could become extinct within a few years.
However, Coxon recently told the Stanford Daily that he might return to Anthropic because he believes he could “make a bigger impact” by working on AI safety there.
Anthropic chief executive Dario Amodei has also identified safety as a major challenge, beyond the data center power constraints that many other AI companies describe as their biggest obstacle. Communities near data center projects have raised concerns about rising electricity costs, transportation and housing pressures, environmental pollution and the limited economic benefits of construction.
Community and environmental opposition
At an Anthropic site being developed by Nscale in West Virginia, heavy rains that coincided with land grading flooded neighboring backyards. “We’re frustrated and angry,” said Matheny, a community spokesperson. Matheny said profit should not be the only consideration and that data centers should be built in more appropriate locations. Nscale says it repaired the damage and bought out some homeowners at prices above market value.
An environmental activist pointed out that Anthropic has made no commitment to using renewable electricity in its data centers and has never published environmental impact reports.
“It’s clear that it would be extremely difficult to explicitly authorize the development of new data centers in the United States,” Tungs said. The backlash could push Anthropic and other companies to focus on data centers outside the United States, which Tungs said would increase risk, complexity and cost.
Concerns over Anthropic’s IPO
A number of labor and other organizations, including National Nurses United, have urged retirement funds to avoid investing in initial public offerings, saying they “should not invest in companies that harm our communities and enable a handful of individuals to amass historic wealth and power.”
Yolanda Daniel, a member of Mississippi Environmental Advocates, a coalition of homeowners’ groups, said Anthropic was socializing the costs of AI development while privatizing its profits. “That’s not how public corporations operate,” she said, vowing to do everything in her power to stop Anthropic from opening a data center across the street next year.
Anthropic’s IPO documents argue that some public benefits could come from making the company’s employees and early investors the world’s largest philanthropists. There is no guarantee, however, that this will happen or that donated money would support causes that directly benefit communities affected by the company’s expansion.
The documents also state that if an employee donates stock to one of Anthropic’s philanthropic funds, voting rights would be transferred to Amodei, giving him greater control.
Amodei and Anthropic’s co-founders hold a majority of the company’s shareholder voting rights. Some investor watchdog groups have warned that this represents centralized control. A successful stock market debut could nonetheless place investors’ trust in Amodei and his associates as they attempt to build the world’s largest company—if Anthropic can secure enough employees and data centers to support that growth.
Source: www.wired.com












