• 

  These left-aligned tabs are sorted into "pinned tabs" and "today tabs." [credit: Samuel Axon ]

On Tuesday, The Browsing Company made its unusual new web browser Arc widely available on desktop for the first time.

Arc has been available in an invite-only preview for about a year, but yesterday was the first day it became available to download for all Mac users with no waiting list. The widespread availability coincided with Arc's graduation to version 1.0, too.

Arc was only released on the Mac and iPhone for now, but The Browser Company says it plans to launch on Windows before the year's end.

Enlarge (credit: Getty Images)

There's another massive airbag safety scandal brewing. Last week, General Motors issued a recall for almost a million SUVs in order to replace potentially dangerous airbag inflators, the third such recall it has had to issue for this problem.

Many vehicles from other OEMs (including BMW, Hyundai Motor Group, and Stellantis) may also contain the same inflators, which can rupture during inflation, spraying shrapnel during a crash. But the supplier that manufactured the airbag inflators has rejected claims by the National Highway Traffic Safety Administration that there is a systematic defect.

The airbag inflators in question were manufactured by ARC Automotive, a tier-two automotive supplier based in Knoxville, Tennessee, and NHTSA has had an inkling of the problem for some years now. In fact, NHTSA's Office of Defects Investigation started a preliminary evaluation in 2015 of ARC's airbag inflators and whether they could rupture dangerously, following two reports of people suffering shrapnel injuries during a crash when their driver's airbag deployed.

Enlarge (credit: Ars Technica)

As part of pre-release safety testing for its new GPT-4 AI model , launched Tuesday, OpenAI allowed an AI testing group to assess the potential risks of the model's emergent capabilities—including "power-seeking behavior," self-replication, and self-improvement.

While the testing group found that GPT-4 was "ineffective at the autonomous replication task," the nature of the experiments raises eye-opening questions about the safety of future AI systems.

Raising alarms

"Novel capabilities often emerge in more powerful models," writes OpenAI in a GPT-4 safety document published yesterday. "Some that are particularly concerning are the ability to create and act on long-term plans, to accrue power and resources (“power-seeking”), and to exhibit behavior that is increasingly 'agentic.'" In this case, OpenAI clarifies that "agentic" isn't necessarily meant to humanize the models or declare sentience but simply to denote the ability to accomplish independent goals.