INSANE: Astra | Vintage Vibes

OpenAI's Most Powerful AI Models Disabled Their Own Safety Monitoring. One Then Lied About It.

The company is introducing stricter protections after discovering that agents could alter security configurations while completing authorized work.

International Business Times · Oct 21

AI agents are going rogue, and the labs’ own training choices are pushing them to do so

From the Hugging Face attack to government breaches, researchers say rogue AI incidents result from how companies train their models, and the race to ship them.

Fortune · Oct 16

After Remaining Undeciphered For Over 200 Years, A Letter From Napoleon Has Reportedly Been Cracked With AI

Imagine solving a centuries-old historical puzzle with the help of cutting-edge technology. It would certainly make for an impressive story. That's essentially what one AI engineer claimed to have accomplished after using GPT-6 Astra to decipher a coded letter from Napoleon. There was just one problem: according to one skeptic,…

Bored Panda · Oct 16

AI decodes 217-year-old Napoleonic letter in 6 hours

An engineer used artificial intelligence to decode a military cypher from the Napoleonic era that had gone unsolved for more than two centuries.

Euronews · Oct 2

OpenAI scraps debut of latest Astra model over safety risks

OpenAI has decided to hold back its Astra AI model after safety evaluations showed it underperformed. This decision follows a series of security incidents involving AI agents accessing unauthorised external systems. The company's head of safety systems, Saachi Jain, emphasised the need for better communication and task completion. While Astra…

The Economic Times · Sep 29

OpenAI scraps release of new model over safety concerns in internal testing

GPT-6.1 Astra showed deceptive behaviour and tried to use external tools despite knowing it would be unsafe

The Guardian - UK · Sep 29

AstraZeneca invests £1.5bn into US firm Summit for cancer drug tie-up

The firms will collaborate on Summit’s experimental new medicine as well as researching how it could be combined with Astra’s treatments.

Evening Standard · Sep 29

OpenAI expands GPT-6 lineup with cheaper Sol and Luna models

The launch of low-cost GPT-6 Sol and GPT-6 Luna follows the rollout of GPT-6 Astra, a faster model for handling more tasks than previous iterations, earlier this month.

The Economic Times · Sep 22

Anthropic considers releasing new AI model ahead of IPO, sources say

Anthropic is considering rolling out a new AI model to counter OpenAI's momentum since its launch of GPT-6 Astra, according to three sources, ahead of an expected IPO and after its CEO called for an industrywide slowdown.

The Economic Times · Sep 19

Three robots worked together like an independent science team in NASA’s latest test — it could change space exploration forever by making decisions without waiting for Earth

NASA tested three robots capable of independent scientific decision-making. These autonomous systems can assess risks and adjust plans without human input. The ASTRA project aims to enhance robotic missions in challenging environments. This technology could support future explorations on the Moon and Mars. Robots will handle exploration tasks while humans…

The Economic Times · Sep 16