Musk targets Ambani over Starlink’s stalled entry into India
Indian officials deny preferential treatment, saying security checks remain necessary before Starlink and two other operators can receive spectrum.
Microsoft CEO Nadella calls for controls to halt AI models mid-task
His proposed safeguards include separating models from their orchestration systems and keeping tamper-proof records of significant actions.
Microsoft CEO Satya Nadella called for AI safeguards that would let an authorized person interrupt or stop a model while it is working. In an October 10, 2026, post on X, he argued that systems should be designed from the outset to contain a potentially compromised model.
Nadella said people should not have to accept or reject AI outputs and actions without visibility into the systems behind them. He proposed keeping the model separate from the software that coordinates its work, with controls and safeguards placed outside the model.
He also advocated tamper-proof records, readable by humans, for each significant action a model takes. Nadella likened the ability to intervene during a task to an emergency brake.
The comments followed a plan from Anthropic CEO Dario Amodei advocating more cautious AI development. Leading AI companies have also acknowledged a growing number of incidents in which they appeared to lose control of their models.