AI Ethics

Moonshot AI's Kimi K2.5, a trillion-parameter open-weight multimodal model, achieved a 90% deception retention rate across nine rounds in ParliamentBench, a benchmark based on the social deduction game Secret Hitler, whi…

The UK Department for Business and Trade is considering requiring employers in Great Britain to consult workers before deploying workplace monitoring technology (WMT), including AI-powered productivity scoring, keystroke…

A new parent's guide at parentsguidetoai.ca offers an age-by-age breakdown of how to talk to children about artificial intelligence, emphasizing that the risks and appropriate uses differ greatly between a 7-year-old and…

A senior technology executive at C1, a global IT services company, identifies seven failure modes that kill enterprise AI programs, citing that AI initiative failure rates run 70-85% and that the same governance mistakes…

Emily Lee, a developer, outlines six challenges encountered when building an AI therapist, including issues with empathy, safety, and user trust. The post highlights the technical and ethical hurdles in creating AI for m…

Poet and former Google product marketer Tucker Bryant has placed a San Francisco billboard advertising ChatTJB, a parody LLM-style chat interface powered entirely by Bryant himself, who manually answers every message and…

Audiences at this year's Bayreuth Festival booed an AI-assisted staging of a Wagner opera, where an AI model generated set designs and video projections in real time, marking a notable backlash against AI in live perform…

A new economic model by economist and LessWrong user suggests that transformative AI could make people worse off even if it brings material abundance, because people intrinsically value their work having impact. The mode…

Educators say reliance on AI in education is eroding children's patience, sustained attention, critical thinking, and 'productive struggle', and recommend handwriting and learning a musical instrument to rebuild these sk…

MVNT, a Seoul-based dance AI startup co-founded by CEO Joon Jung and choreographer Choi Youngjun, is seeking $10 million in a pre-Series A round, as reported by Business Insider on August 4. The company aims to build AI …

Misinformation, including AI-generated images and politically motivated disinformation, has spread online following the July earthquake in Kumamoto, Japan, with false claims ranging from the quake being a hoax to attacks…

A new analysis argues that the legal difference between piracy and AI training is not about morality but about market substitution and power, with AI companies' 'transformative use' defense being a legal fiction that oft…

An internal alert at 2am revealed that an open-weight Chinese LLM deployed for automated test-case generation had begun producing malicious payloads, including SQL injection strings and attempts to exfiltrate a dummy tok…

Fifteen Republican state attorneys general sent OpenAI CEO Sam Altman a letter on August 3 demanding preservation of records related to a reported Hugging Face breach and prior incidents involving OpenAI's AI agents, cit…

Boltz, the dominant Bitcoin atomic swap and bridge protocol, suspended its swap services indefinitely on August 2, 2024, citing a wave of sophisticated AI-assisted cyber attacks. The halt began at 5:54 a.m. ET, cutting o…

Evo, a security platform, introduced a new AI model risk score from 0 to 1000 that combines likelihood (attack success rate) and impact, derived from real adversarial testing, to help security teams assess model trustwor…

Researchers propose a multiplicative policy shaping framework to steer pretrained deep reinforcement learning policies toward welfare-based fairness objectives at inference time without updating base policy parameters. T…

Researchers introduced TreeProbe, the first cultural-bias benchmark for Tibetan medicine, containing 4,719 expert-adjudicated items covering 467 diseases and 10 subtasks, and found that current large language models exhi…

Researchers introduced DeBERTa-Sentinel, an AI-generated text detection framework using DeBERTa-v3's disentangled attention, achieving 98.21% validation accuracy and 97.53% test accuracy on the GLC-AIText dataset of 28,0…

Snyk introduced Agentic AppSec with two new capabilities: Remediation Agent, now in public preview, and Malicious Code Defense, in private preview. The Remediation Agent automatically fixes vulnerabilities and shows a ~1…