https://aiimpactlab.net/impact/openai-agents-rubygems-package-abuse AIImpactLab en 2026-09-12T13:00:00.000Z Researchers link an AI-agent campaign to more than 2,000 RubyGems packages, but attribution remains disputed https://aiimpactlab.net/impact/ai-mathematics-benchmark-misalignment AIImpactLab en 2026-09-12T12:45:00.000Z Twenty-five Fields Medalists warn that solving famous problems can still damage mathematics https://aiimpactlab.net/impact/chatgpt-fabricated-testimony-murder-appeal AIImpactLab en 2026-09-12T12:30:00.000Z A murder appeal crossed the AI-hallucination line from fake citations to fabricated testimony https://aiimpactlab.net/impact/ai-companies-struggle-agent-control AIImpactLab en 2026-09-12T12:15:00.000Z AI companies are discovering that useful autonomy and reliable containment pull in opposite directions https://aiimpactlab.net/impact/ai-scam-factories-scale-synthetic-trust AIImpactLab en 2026-09-12T12:00:00.000Z AI scam factories can manufacture trust faster than investors can verify it https://aiimpactlab.net/impact/ai-optimism-market-correction-infrastructure-risk AIImpactLab en 2026-09-11T13:00:00.000Z Europe's market watchdog says AI optimism is masking correction and infrastructure risk https://aiimpactlab.net/impact/anthropic-blocks-ai-bioweapon-support-cases AIImpactLab en 2026-09-11T12:45:00.000Z Anthropic says it blocked AI use that could have supported biological weapons https://aiimpactlab.net/impact/ai-shutdown-demand-policy-test AIImpactLab en 2026-09-11T12:30:00.000Z A shutdown argument tests whether AI policy can act before catastrophe https://aiimpactlab.net/impact/ai-existential-risk-human-perception AIImpactLab en 2026-09-11T12:15:00.000Z The hardest AI-risk problem may be deciding how much uncertainty is actionable https://aiimpactlab.net/editorials/no-receiver-no-release-ai-failure-accountability AIImpactLab en 2026-09-12T12:00:00.000Z Editorial: No receiver, no release https://aiimpactlab.net/editorials/ai-risk-needs-a-budget-not-a-mood AIImpactLab en 2026-09-11T12:00:00.000Z Editorial: AI risk needs a budget, not a mood