The AI Boom’s Hidden Costs

The Costs of the AI Boom Come Into View

For much of the past two years, the public conversation around artificial intelligence has been dominated by soaring valuations, blockbuster funding rounds and the race among technology companies to build ever more powerful systems. A new set of developments is redirecting that focus toward a different question: who bears the costs when those systems spread.

In recent days, concerns about artificial intelligence have converged around three fronts that are less about invention than impact — jobs, safety and corporate responsibility. New reporting and labor research suggest that women may be especially vulnerable to workplace disruption from generative AI. Anthropic, one of the leading AI start-ups, has said it will set aside a slice of its own equity for a public-benefit fund that could help address such disruption. At the same time, prosecutors and watchdog groups are highlighting risks of misuse, from an espionage case involving OpenAI to the ease with which teenagers can still obtain apps used to create nonconsensual sexualized images.

Taken together, the episodes underscore how the AI debate is evolving. The issue is no longer only whether the technology works, or how much money it can attract, but whether companies and regulators are prepared for the social consequences arriving alongside it.

A Gendered Risk in the Labor Market

Among the clearest warnings comes from the labor market. Recent analysis, drawing on findings from the International Labor Organization, suggests that women are more exposed than men to AI-related disruption because the kinds of jobs generative AI can most readily reshape — clerical, administrative and other office-based roles — are disproportionately held by women.

The ILO said in March that female-dominated occupations are almost twice as likely as male-dominated ones to be exposed to generative AI. In high-income countries, it found, the occupations at highest risk account for roughly 9.6 percent of women’s employment, compared with 3.5 percent of men’s.

That does not necessarily mean a one-for-one wave of layoffs. Economists have repeatedly cautioned that automation often changes tasks before it eliminates jobs altogether. But the historical pattern is sobering. Past technological shifts, including the decline of telephone switchboard operators and the spread of personal computers through administrative work, often fell hardest on roles filled largely by women. AI may now be poised to accelerate a similar pattern, especially in white-collar settings that had once seemed relatively insulated from automation.

Why that matters extends beyond the labor market. If companies use AI chiefly to strip out routine administrative work without comparable investments in retraining, job redesign or internal mobility, the technology could widen existing gender inequalities in pay, advancement and job security.

A Company Tries to Answer the Question of Remedy

That prospect helps explain why Anthropic’s announcement resonated beyond the usual Silicon Valley fanfare. The company said on Oct. 18 that it would dedicate 5 percent of its shares to a new public-benefit fund that could be used, among other things, to address labor disruption linked to AI.

Anthropic said the fund would be established outside the company as a limited liability company, a structure it said would give it flexibility in how the money is deployed. The company also said that, combined with earlier arrangements, about 20 percent of its shares are now effectively earmarked for public-good purposes.

The announcement was notable not simply for its size, which could become substantial if Anthropic’s valuation continues to rise, but for the premise behind it. AI companies have generally argued that their products will drive broad productivity gains and eventually create more opportunity than they displace. Anthropic’s move suggests a more explicit recognition that the companies building these systems may also need to finance remedies for harms that accompany them.

Significant questions remain. Anthropic has not yet detailed who will govern the fund, how quickly it could deploy money or what portion would be directed specifically toward workers affected by automation. Still, the move marks a shift in tone. In the emerging politics of AI, it may no longer be enough for companies to promise benefits in the abstract; they may also be expected to help shoulder the costs.

Security Fears Reach the AI Lab

Elsewhere, attention has turned to the vulnerabilities inside the companies developing frontier AI.

Federal prosecutors have expanded the case against Abdulrahman Alfozan, a Saudi national accused of planting a recording device in an executive conference room at OpenAI’s headquarters in San Francisco. According to prosecutors, a hidden recorder was discovered on Sept. 30, and another device was later found in Alfozan’s bag. Court filings now also say he had frequent WhatsApp contact with a Saudi royal in the days before his Oct. 9 arrest.

Prosecutors are arguing that he should remain detained. The public record has not established who, if anyone, directed the alleged spying, and the case has not been adjudicated. But the allegations have sharpened concerns that AI labs are becoming targets not just of commercial competition, but of possible foreign intelligence interest.

That matters because the most advanced AI systems are increasingly treated as strategic assets. As governments and corporations race to secure talent, chips and proprietary research, the security perimeter around AI development is beginning to look more like that of a sensitive defense contractor than a conventional software company. The OpenAI case, whatever its eventual outcome, highlights how quickly those concerns have moved from theory to courtroom.

A Child-Safety Gap in Plain Sight

At the consumer end of the AI ecosystem, a separate investigation has raised alarms about how easily minors can still access tools designed for abuse.

A report by the Tech Transparency Project found that test accounts set up to resemble 13-year-olds were able to download 135 apps with features to generate nonconsensual nude imagery, despite parental controls offered by Apple and Google. The findings suggest that app-store safeguards and age restrictions may be failing to block software that can be used to create sexually explicit fake images of real people.

The stakes are not hypothetical. Child-safety advocates have spent much of the year warning that AI has lowered the barriers to producing image-based sexual abuse material, including fabricated explicit images made from ordinary photos. Cases involving teenagers have brought the issue into schools and families with particular force, where the harms can be immediate, personal and difficult to reverse once content begins circulating.

The “nudify” apps identified in the investigation sit at the intersection of several unresolved problems: weak app-store enforcement, patchy age verification and the speed with which generative AI tools can be repackaged for exploitation. Regulators have begun paying closer attention to AI-generated sexual abuse material, but enforcement remains uneven, and it is unclear how quickly major platform operators will remove or reclassify offending apps.

The Debate Moves From Possibility to Accountability

None of these developments resolves the central uncertainties surrounding AI. It is still unclear how quickly changes in office work will translate into actual job losses, or whether the main effect will be task reshaping rather than mass displacement. It is not yet known how Anthropic’s fund will be run or whether it will become a model for rivals. The OpenAI case has yet to be tested in court. And Apple and Google may yet tighten controls in response to the findings on nudify apps.

But the broader direction is becoming harder to miss. As AI systems move from novelty to infrastructure, the fallout is becoming more concrete and more unevenly distributed. Women in clerical work may face greater employment risk. Children and teenagers may encounter tools built for humiliation and abuse. Companies creating advanced AI may find themselves confronting security threats as well as ethical ones. And the public is beginning to ask not only what these systems can do, but who is responsible when they do harm.

That shift may prove to be one of the most important changes in the AI era so far. The technology industry has spent years making the case for acceleration. It is now being pressed to make a case for accountability.