Tech News
Back to home
Technology

OpenAI’s “compromise” with the Pentagon is what Anthropic feared

Source

MIT Technology Review

Published

TL;DR

AI Generated

OpenAI recently announced a deal allowing the US military to use its technologies in classified settings, following a public reprimand of Anthropic by the Pentagon. OpenAI emphasized that the agreement protects against autonomous weapons and mass surveillance. While OpenAI claims to embed safety rules into its models for the military, it remains unclear how these rules differ from those for normal users. The company's approach differs from Anthropic's focus on specific prohibitions, instead citing laws and policies as its basis for working with the Pentagon. There are concerns about whether OpenAI's compromise will satisfy critical employees and how the Pentagon will transition from using Anthropic's AI model to OpenAI's amid escalating tensions in the Middle East.

Read Full Article

Similar Articles

MIT Technology Review

The Download: murderous ‘mirror’ bacteria, and Chinese workers fighting AI doubles

Scientists initially proposed creating mirror bacteria with reversed proteins and sugars for research purposes, but now fear these organisms could pose a catastrophic threat to life on Earth. In China, tech workers are concerned about AI replication of their skills and personalities, leading to fears of losing professional identity. Meanwhile, the White House and Anthropic are working on a compromise, and other tech news includes Palantir's manifesto, Germany's push for looser AI rules, and Nvidia's shift towards AI over gaming.

MIT Technology Review•
Bluetooth tracker hidden in a postcard and mailed to a warship exposed its location — $5 gadget put a $585 million Dutch ship at risk for 24 hours

Bluetooth tracker hidden in a postcard and mailed to a warship exposed its location — $5 gadget put a $585 million Dutch ship at risk for 24 hours

A Dutch warship, HNLMS Evertsen, part of a NATO carrier strike group, unknowingly exposed its location for 24 hours after receiving a postcard with a hidden Bluetooth tracker. The tracker, similar to an Apple AirTag but cheaper at $5, was mailed in, allowing spies to track the $585 million Navy ship without physical access. Dutch authorities discovered and disabled the tracker within a day, prompting a ban on electronic greeting cards due to security risks. This incident highlights the ongoing challenges of maintaining operational security in the face of new technologies that can inadvertently compromise military operations.

Tom's Hardware•
News outlets are blocking Wayback Machine from archiving their pages — 23 outlets concerned AI companies might abuse fair use and use it to train their models

News outlets are blocking Wayback Machine from archiving their pages — 23 outlets concerned AI companies might abuse fair use and use it to train their models

Many news outlets are blocking the Wayback Machine from archiving their pages due to concerns that AI companies may misuse fair use policies by using archived content to train their models. This move could limit access to historical news stories and crucial information in an era where misinformation is rampant and AI models can generate convincing but false answers. Notably, 23 major publications, including USA Today and The New York Times, are currently blocking the Internet Archive's crawler. While some argue that publications should handle their own archiving, a neutral third party like the Wayback Machine is crucial for maintaining accurate historical records and tracking changes in online content. Despite concerns about fair use, preventing archiving services like the Wayback Machine could have more negative consequences than benefits for society.

Tom's Hardware•
UK confirms drone-killing DragonFire laser weapon for Royal Navy destroyers by 2027 —laser downs 400mph high‑speed drones, costs $13 per shot

UK confirms drone-killing DragonFire laser weapon for Royal Navy destroyers by 2027 —laser downs 400mph high‑speed drones, costs $13 per shot

The UK Ministry of Defence has confirmed the deployment of the DragonFire laser weapon on Royal Navy destroyers by 2027, five years earlier than planned. The laser, developed by MBDA UK, can take down high-speed drones traveling at 400mph and costs $13 per shot. This system is more cost-effective compared to traditional interceptor missiles. Trials have shown successful drone interceptions, prompting an accelerated timeline for the program. The MoD is also considering land and air applications for the DragonFire technology.

Tom's Hardware•

We use cookies

We use cookies to ensure you get the best experience on our website. For more information on how we use cookies, please see our cookie policy.