Top 10 Ars Technica — Latest — April 10, 2026

Anthropic ran Claude through 20 hours of simulated psychiatric sessions, attempting to build an AI with something resembling emotional stability. The novelty is less the therapy and more the admission that the largest models are psychologically brittle.

A clinical trial proved that CRISPR-based gene editing can treat β-Thalassaemia, adding another blood disorder to the list of conditions curable by design. This moves the therapy closer to becoming a standard option rather than a desperate last resort.

A rival virtualization company claims that anger over Broadcom's post-acquisition pricing and support is driving a mass exodus from VMware. The 'thousands of migrations' figure suggests Broadcom's win-the-then-worry-later strategy is backfiring.

A Ugandan chimpanzee troop splintered into two factions, and the resulting conflict ended with one group killing rivals. The observation is a stark reminder that intergroup violence predates human politics by millions of years.

The four Artemis II astronauts, slated to fly around the Moon, admitted that the scale of their mission hasn't fully hit them yet. Their candor is refreshing—spaceflight isn't routine, and pretending otherwise does no one any favors.

Trump-appointed federal judges refused to block Trump's executive order blacklisting Anthropic's AI technology, upholding a rare instance of the administration targeting a specific company. The ruling leaves Anthropic's U.S. operations in legal limbo.

Volkswagen halted U.S. production of its ID.4 electric SUV, claiming it has enough inventory to last into 2027. For an automaker that bet billions on EVs, this is less a pause and more a white flag on demand.

A Texas police corporal used driver's license photos from traffic stops to generate AI-generated child sexual abuse material. The case is a horrifying example of how law enforcement tools become weapons when trust is absent.

The first person convicted under the new Take It Down Act—which criminalizes non-consensual intimate images—continued generating AI nudes even after his arrest. The law's existence didn't stop him, raising hard questions about deterrence in the AI era.

Anthropic ran Claude through 20 hours of simulated psychiatric sessions, attempting to build an AI with something resembling emotional stability. The novelty is less the therapy and more the admission that the largest models are psychologically brittle.

A clinical trial proved that CRISPR-based gene editing can treat β-Thalassaemia, adding another blood disorder to the list of conditions curable by design. This moves the therapy closer to becoming a standard option rather than a desperate last resort.

A rival virtualization company claims that anger over Broadcom's post-acquisition pricing and support is driving a mass exodus from VMware. The 'thousands of migrations' figure suggests Broadcom's win-the-then-worry-later strategy is backfiring.

A Ugandan chimpanzee troop splintered into two factions, and the resulting conflict ended with one group killing rivals. The observation is a stark reminder that intergroup violence predates human politics by millions of years.

The four Artemis II astronauts, slated to fly around the Moon, admitted that the scale of their mission hasn't fully hit them yet. Their candor is refreshing—spaceflight isn't routine, and pretending otherwise does no one any favors.

Trump-appointed federal judges refused to block Trump's executive order blacklisting Anthropic's AI technology, upholding a rare instance of the administration targeting a specific company. The ruling leaves Anthropic's U.S. operations in legal limbo.

Volkswagen halted U.S. production of its ID.4 electric SUV, claiming it has enough inventory to last into 2027. For an automaker that bet billions on EVs, this is less a pause and more a white flag on demand.

A Texas police corporal used driver's license photos from traffic stops to generate AI-generated child sexual abuse material. The case is a horrifying example of how law enforcement tools become weapons when trust is absent.

The first person convicted under the new Take It Down Act—which criminalizes non-consensual intimate images—continued generating AI nudes even after his arrest. The law's existence didn't stop him, raising hard questions about deterrence in the AI era.