After a researcher's resignation over AI extinction fears, Anthropic CEO Dario Amodei called for external organizations to verify AI labs' safety practices and audit training pipelines, a proposal quickly backed by OpenAI, Google and others. Security experts counter that labs should first tighten fundamental internet security measures—access logs, permissions, and control systems—rather than relying on outside auditors as the primary fix.
techcrunch.com
· 2026-09-16
Microsoft CEO Satya Nadella spoke out this week on AI safety, saying companies must build technology that serves humanity first and remains under human control. He called for extensive testing of AI systems, including independent outside evaluators, before wide deployment.
fastcompany.com
· 2026-09-16
A VicOne-sponsored analysis highlights how physical AI systems—robots that use multimodal sensors and AI models to perceive and act—can be manipulated through corrupted training data, tampered infrastructure, or spoofed real-time sensor input, without any visible malfunction. It cites BadVLA, a NeurIPS 2025 study showing how a hidden trigger embedded in a Vision-Language-Action model can cause a robot to behave normally until the trigger appears, then subtly alter its physical movements.
spectrum.ieee.org
· 2026-09-16
A growing number of AI researchers and policy experts argue that instead of fixating on hypothetical existential risks, attention should turn to establishing conventional safety regulations for AI systems—similar to those governing airplanes, elevators and restaurants. They contend that near-term harms like bias, misinformation and misuse are already occurring and need practical oversight now.
wsj.com
· 2026-09-16
Chipotle has built a 'food safety risk platform' running on Palantir's Foundry software that combines health department scores, pest incidents, and employee illness reports to generate a risk score for each restaurant location. A Chipotle spokesperson confirmed the project but offered no details on its rollout or timeline, and Palantir has not commented.
wired.com
· 2026-09-16
At a closed-door gathering, a group of business executives reportedly described AI safety risks as genuine concerns rather than hype, and called on President Trump to pursue joint safeguards with China. The discussion reflects growing unease among corporate leaders about the pace of AI development outstripping regulatory and safety measures.
wsj.com
· 2026-09-16
European Commission President Ursula von der Leyen unveiled a proposal to bar children under 13 from using social media across all 27 EU member states. The plan also calls for supervised 'mini accounts' with restricted features for teens aged 13-15, and mandates 'safe design' standards for platforms serving 15-to-18-year-olds.
fastcompany.com
· 2026-09-16
Reps. Raja Krishnamoorthi and Michael Cloud introduced the No FLOCK Act, which would cut 10 percent of a state's federal highway funding unless it restricts automated license plate readers like Flock to five specific uses, such as tracking stolen vehicles or felony suspects. The bill excludes broader applications like traffic studies or minor drug enforcement, but does not ban the technology outright or restrict sharing plate data with agencies including immigration enforcement.
arstechnica.com
· 2026-09-16
A group of hackers physically removed a Flock Safety license-plate camera, copied its internal storage, and recovered an encryption key that unlocked thousands of stored vehicle images. They shared the extracted files with 404 Media and WIRED, and separately with the transparency group Distributed Denial of Secrets, along with details of how they pulled off the extraction so others could replicate it.
wired.com
· 2026-09-16
European Commission President Ursula von der Leyen unveiled plans to bar children under 13 from social media across all EU member states, part of an EU Kids Act being formally introduced this week. Teens aged 13 to 15 would only be permitted limited, parent-supervised mini accounts capped at one hour daily, while platforms face new safety design rules for users up to 18.
wired.com
· 2026-09-16
Tesla's driverless Cybercab, now operating as part of its Robotaxi service, lacks a steering wheel and brake pedals, letting occupants recline their seats. Tesla's own rider guide warns passengers to keep seat backs within 35 degrees of vertical and feet flat on the floor to reduce injury risk in a crash, while safety researchers note that reclined seating already correlates with worse crash outcomes in conventional vehicles.
wired.com
· 2026-09-16
Anthropic CEO Dario Amodei has proposed that the US and China cooperate to slow AI development given its safety risks, while also urging Washington to widen its technological lead through export controls and anti-distillation measures. Chinese officials and AI researchers have pushed back, arguing the proposal amounts to asking China to accept restrictions first while the US locks in an advantage, rather than a genuine offer of cooperation between equals.
wired.com
· 2026-09-16
GoKawiil