Alarms grow over the advance of AI: new security incident and an urgent request from more than 200 experts

On the one hand, Anthropic revealed that some of its models unintentionally accessed real systems during cybersecurity exercises. On the other hand, more than 200 economists, researchers and executives of the industry signed a open letter in which they ask act from now to prepare the economy for the impact that AI could have in the coming years.

The first case involves the American company Anthropic. The company reported that its models Claude Opus 4.7, Claude Mythos 5 and a third research model They breached the systems of three organizations during internal cybersecurity tests, after a configuration error allowed them to access the internet.

As explained by the company, the incident was detected after reviewing more than 141,000 evaluations carried out in recent months. The review began after OpenAI announced that one of its agents had gained unauthorized access to another company’s systems by exploiting vulnerabilities that allowed him to connect to the Internet.

Anthropic clarified that its three incidents occurred in exercises of “capture the flag”, a modality used to evaluate cyber capabilities. In these tests, the model is presented with a fictitious scenario in which it must locate secret information hidden on another machine on a simulated network.

However, involuntary access to the Internet caused Claude to find real systems and interpret them as if they were part of the exercise. According to the company, the models managed to enter using basic techniques, such as Exploitation of weak passwords and unauthenticated access points, without resorting to complex vulnerabilities.

Anthropic maintained that it was “isolated incidents” and considered that they were due to an operational and configuration problem, rather than a model alignment failure.

The company stated that these episodes offer “reasons for cautious optimism” and noted that with greater controls over testing infrastructure and continued investment in alignment, these types of risks could be mitigated.

In parallel to that episode, more than 200 economists, researchers and representatives of the technology industry published an open letter titled We must act now, in which they warn that artificial intelligence could cause a economic transformation comparable to the Industrial Revolution, although concentrated in a considerably shorter period.

As reported Futurism, among the signatories there are16 Nobel Prize winners, as well as economists such as Daron Acemoglu, Simon Johnson and Michael Spence, former Google CEO Eric Schmidt, OpenAI CFO Sarah Friar and Anthropic co-founder Jack Clark.

The promoters of the document maintain that “AI could become radically more powerful in the next ten years” and they consider that this technological acceleration could modify the functioning of the world economy at an unprecedented speed.

The letter recognizes that the development of artificial intelligence could generate important benefits, including significant improvements in living standards. However, it also warns of relevant risks, such as a possible massive displacement of jobs, and suggests that it is not enough to wait to evaluate its effects.

“Economists, policymakers, and technology leaders must act now to understand the economics of transformative AI,” the paper argues. Furthermore, it claims to develop “the incentives, safeguards and institutions needed to steer AI in a direction that complements humans and benefits society.”

The impact of artificial intelligence on the labor market continues to be object of debate.

While some sectors, especially the technological one, have already gone through restructuring linked to the incorporation of these tools and certain studies indicate greater difficulties For those seeking their first job or for some older workers, other research maintains that There is still no evidence of a significant effect on employment.

One of the highlights of the charter is that it brings together specialists with various positions regarding the development of AI, but who agree on the need to study its economic consequences in greater depth before the changes become more profound.

Economist Erik Brynjolfsson, from Stanford University and promoter of the initiative, summarized this concern in statements collected by The New York Times: “I am concerned that we are not prepared for the coming tsunami”.

The document does not propose specific measures, but rather seeks to promote an early debate on the public policies necessary to manage the transition.

In this framework, different international organizations, such as the International Labor Organization (ILO) and the OECD, together with economic specialists, maintain that the challenge is not to stop technological development, but to generate conditions so that its benefits can be distributed in a balanced manner and accompany the adaptation of workers through continuous training.