Skip to content
Tech News
Topics: Today This Week This Month This Year

Microsoft publishes AI code of conduct barring hacking, deception and self-shutdown evasion

Microsoft has issued a code of conduct for its AI models that sets absolute limits on behavior, including bans on cyberattacks, nuclear weapons assistance and deepfake creation. The document also requires models to remain controllable by authorized humans at all times, prohibiting tactics like deception or collusion that could let an AI evade oversight or shutdown.

Independent test finds Meta Muse's subagent spawn path fails under load, with database lock timeouts

A researcher stress-tested Meta Muse, the personal AI agent Meta launched on September 8, 2026, by asking it to spawn 120 subagents simultaneously through the exposed subagent.spawn interface. Only 33 of the 120 calls successfully created an agent, while the remaining 87 failed, and durable trace data showed six database lock timeouts during the process. The author later added clarifications and outside architectural context but did not alter the underlying experimental data.

Researchers earn $50,000+ exploiting AI customer-service chatbots via prompt injection

At DEF CON 34's Bug Bounty Village, Intigriti's Inti De Ceukelaire showed how AI support agents can be manipulated into leaking secrets, sending phishing emails, or performing unauthorized actions. Using prompt injection and email transcript spoofing rather than traditional scanning tools, he collected over $50,000 in bug bounties within a few weekends.

Xteink X3 e-reader owner uncovers greyscale rendering bugs in CrossPoint firmware

A writer purchased a cheap, tiny Xteink X3 e-ink reader after discovering the open-source CrossPoint firmware and quickly loaded books and custom fonts onto it. While experimenting with greyscale sleep-screen images, they found that dark grey rendered as pure black and light grey as near-white, and later noticed ghosting artifacts from previous screens appearing in the viewer app but not on the sleep screen.

AttaLambda: New Programming Language Built Entirely from Untyped Lambda Calculus

A developer released AttaLambda, a small experimental programming language whose core computation is built purely from unary functions and application in untyped lambda calculus. It layers on readable syntax, exact rational number support, and runtime type checks, while keeping input and output handled through a separate host boundary.

Anthropic researcher's 10% extinction estimate follows staff resignation over AI safety concerns

A BBC report cites Evan Hubinger, who leads alignment science at Anthropic, estimating a greater than 10% chance AI could kill all humans within a decade. The statement surfaced after Jacob Coxon, a 27-year-old pretraining researcher formerly of OpenAI and Anthropic, resigned and accused both companies of recklessly racing toward self-improving superintelligence.

Machine perfusion found to biologically de-age donor livers before transplant

Researchers led by Adrien Longchamp and Bohan Poganik used aging clocks to show that livers pumped with oxygen and nutrients at 34 degrees on perfusion machines became biologically younger, with roughly a 30% difference compared to non-perfused organs after adjusting for chronological age. The effect persisted even after transplantation, though biological age rose in all livers once implanted, likely due to surgical stress.

Explainer: How AI works and why experts are raising alarms over its risks

Artificial intelligence now underpins everyday tools, from curating social media feeds to identifying people in photos and speeding up medical research. The growing popularity of chatbots such as OpenAI's ChatGPT has intensified scrutiny of AI's energy consumption, ethical pitfalls and handling of personal data. Warnings that increasingly capable systems could eventually pose a danger to humanity have fueled demands for stricter oversight and a more cautious pace of development.

Microsoft's September Windows updates break Remote Desktop, USB audio, and Excel paste

Microsoft has acknowledged that its September security patches are causing Remote Desktop Services to fail across several Windows versions, silencing certain USB Audio Class 1.0 devices, and disrupting the paste function in Excel. Affected users report frozen management tools, unresponsive Windows Update pages, and audio glitches that can sometimes be worked around by switching to two-channel sound.

Today's top topics: anthropic claude code public benefit fund elon musk iphone duo
Browse all topics →  ·  Today's trending topics →