“algorithm study” — 23 distilled results

SkillBloat: Token Amplification Attacks via Skill Injection in LLM Coding Agents

21929v1 Announce Type: new Abstract: Agent skills extend coding agents with task-specific instructions, scripts, and resources, but they also create a trusted instruction channel that can be abused beyond conventional security attacks. This paper studies token amplification through skill injection: an economic resource-abuse threat in which a malicious skill causes an agent to consume substantially more tokens than needed for normal task execution.

rss:arxiv-qfin 5d ago · rss:arxiv-cscr 14d ago · rss:arxiv-cscr 6d ago · rss:arxiv-qfin 6d ago · rss:arxiv-cscr 15d ago · rss:arxiv-cscr 12d ago

Mixed news spanning Pluto reclassification, AI, and labor trends

Unrelated collection of stories from August 24–November 17, 2026: commemoration of Pluto's 2006 demotion, SEC investigation of an AI hedge fund, instant ramen history, security updates, Biden immigration's economic impact, RDP honeypot data, and AI adoption in labor markets. No coherent single story.

rss:arxiv-cscr 32d ago · rss:arxiv-qfin 4d ago · rss:arxiv-cscr 28d ago · rss:arxiv-qfin 26d ago · rss:arxiv-cscr 12d ago · rss:arxiv-cscr 7d ago

Samsung Galaxy S27 redesign rumors and technology updates

Leaked CAD renders from August 24–25, 2026, suggest Samsung's upcoming Galaxy S27 Ultra will feature a redesigned wide rear camera plateau. Anthropic linked Claude's memory system between chat and Cowork on August 25, 2026. Separate reports cover systemd SSD damage claims, Thai flooding, kelvin wave ocean phenomena, and NRC workforce assessments.

kite:science 31d ago · kite:science 24d ago · rss:axios 21d ago

Tech firms launch coalition against AI cyberattacks

OpenAI led an open letter signed by more than 100 organizations, including Anthropic, Microsoft, Google, Amazon, and others, on August 27–29, 2026, calling for coordinated defense against rising AI-enabled cyberattacks. Nvidia's new safety platform received cautious praise by November 2026, with experts noting no single solution exists to AI security risks.

rss:arxiv-cscr 32h ago · rss:arxiv-cscr 12d ago