Compare full coverage across 19 outlets
Tech11:08 · Jun 14

U.S. Move to Shut Down Anthropic AI Model Raises Cybersecurity Alarm

YnetCenter
Translated & summarized from Ynet by baba
The story · English

An unusual intervention by the U.S. administration forced Anthropic to suspend two of its new AI models, prompting fresh questions about the risks of the next generation of artificial intelligence. In an interview with ynet, Gil Messing, chief of staff at Check Point, said the episode could reshape cyber warfare and warned it is likely only the beginning.

Messing said the case involved two models, “Mythos 5,” which he described as the most powerful AI model in the world, and “Fable 5,” a smaller version with safeguards. According to him, Anthropic released Fable 5 only after adding protections meant to prevent uses such as highly advanced cyberattacks or biological weapon development. He said those protections lasted “exactly three days” before the U.S. government demanded access be restricted to U.S. citizens and users inside the United States, a condition Anthropic could not realistically meet, so it shut both models down. He said this had “never happened before.”

Explaining Washington’s concern, Messing said Mythos was designed to find software security flaws and did so quickly, easily, and at massive scale, including vulnerabilities never seen before. He called a security flaw a software hole that allows intruders in and can enable severe cyberattacks. In his view, once such a model is broadly available, defenders may use it to close gaps, but attackers gain a weapon they never previously had.

Messing said this is a “doomsday scenario,” and argued the U.S. sees a need for protections similar to those used for weapons and even nuclear technology. Asked whether the danger is comparable to an atomic bomb, he replied, “The short answer is yes, or more or less yes,” saying cyber tools of this kind could badly disrupt daily life because so much now depends on technology and internet connectivity, including hospital machines.

He added that attacks using flaws discovered by a model like Mythos could shut down anything connected to the internet, including infrastructure, road transport, airports and credit systems. Messing said the race will continue even if current models are limited, because another company could soon build an even stronger one. He said the weekend’s events were only “the first stage” and warned, “I just hope we do not regret it one day.”

Read the original at Ynet
Full coverage · 10 outlets
63% centerFirst: Calcalist · Jun 13

The same event, reported separately by each outlet. Open a few to compare what different newsrooms emphasize — and what they leave out.

Center 5Right 3Unrated 2
Related stories · 5

Not the same event — other stories that share this one’s people, places, or theme: background, reactions, and follow-ups.

Open the live terminal