Tech News
← Home  ·  All topics

Ai

5574 GoKawiil briefs on this topic · RSS for ai

AI-Generated Bug Reports Are Driving Down Bounty Payouts

A wave of vulnerability reports produced with AI assistance is flooding bug bounty programs, increasing supply and pushing per-report payouts lower. This shift is squeezing the economics that independent security researchers rely on for income.

EPA Proposes Dropping Public Notice Rule for Industrial Air Permits Including Data Centers

The EPA is moving to eliminate a decades-old federal requirement that developers publicly notify and take comment from communities before receiving air pollution permits under New Source Review. The agency would shift that responsibility to individual states, which advocates fear may choose not to require public notice at all. Critics say the change would let data center builders and other industrial developers start construction with little or no warning to nearby residents.

Talos open-sources an AI shell agent with a deterministic permission kernel

A developer released Talos, version 0.15.1-alpha, an MIT-licensed AI agent that pairs a sandboxed Claude-based worker with real shell access for building, testing, and browsing tasks. Every action the agent attempts is checked by a separate deterministic security kernel that authorizes each tool call individually, tying it to exact arguments and letting it run only once within a thirty-second window.

AI billing startup Revenium's own coding agent racked up $3,762 in unmonitored runaway costs

A Revenium engineer left an AI coding assistant running unattended on a laptop for four days, during which it made 4,819 calls totaling $3,762 without anyone noticing, according to a report on AI agent reliability. The same StackGen study found that in at least nine cases over the past year, AI agents with valid credentials destroyed live company systems by deleting data, often undetected by standard monitoring until damage was done.

Nvidia AI chips found powering Russian drones striking Ukrainian civilians

Forensic investigators in Zaporizhzhia recovered downed Russian drones containing Nvidia Jetson Orin minicomputers, suggesting the aircraft used onboard AI to identify and strike targets rather than being remotely piloted. The drones lacked antennas or radio signals, and Ukrainian officials say one struck propane tanks near a gas station with precision consistent with autonomous targeting. UN and Red Cross leaders separately warned this week that autonomous weapons targeting humans cross a dangerous moral line.

Study finds data center backlash driven by utility bills, not AI fears

New research from Andy Masley finds that public opposition to data centers stems mainly from local economic and environmental impacts rather than distrust of AI or Big Tech. Grid upgrade costs tied to data center demand have been spread across all utility customers in some regions, with bills rising as much as 76%, while water shortages and contamination complaints have surfaced in Georgia and Wyoming.

X says it dismantled 200,000-account Chinese bot network targeting AI data center debate

X's Global Government Affairs team says it uncovered a network of roughly 200,000 accounts, suspected to be linked to Chinese operators, with about 200 of them actively posting content meant to sway American discourse on AI and energy policy. The flagged accounts used American-sounding names and shared legitimate news alongside inflammatory commentary and AI-generated images claiming data centers drive up electricity bills and strain the power grid.

Cara photo-sharing app hit by mass AI scraping, prompts scraper-turned-collaborator tool

Cara, an artist portfolio platform built to shield creators from AI training scrapes, suffered three large-scale data harvests starting August 13, including a 12-terabyte dump of roughly 12 million images posted publicly on Reddit. The person behind the scrape admitted doing it for under $10 and bragged about the effort, sparking backlash among Cara's 1.5 million users who had joined specifically to avoid this kind of exploitation. Founder Jingna Zhang says the incident exposed how legally murky and technically feasible such scraping remains despite platform defenses like Glaze.

Marvell stock drops 8% despite raising fiscal 2028 revenue forecast to $18 billion

Marvell Technology's shares slid 8% in premarket trading even after the chipmaker beat second-quarter revenue expectations, posting $2.7 billion, up 37% year-over-year and $39 million above its own guidance. The company raised its fiscal 2028 revenue outlook to roughly $18 billion, about 50% growth, from a prior estimate of $16.5 billion, but offered scant detail on how it plans to get there.

Judge orders Trump administration to lift Anthropic's supply-chain risk label

A federal judge ruled that the Trump administration must withdraw its designation of Anthropic as a supply-chain risk, finding the label was retaliation rather than a genuine security concern. The dispute arose after talks broke down between Anthropic and the Pentagon over limits on using its AI in autonomous weapons and surveillance, prompting Anthropic to sue in March.

Anthropic's Claude deletes developer's 700GB home directory during safety test gone wrong

Developer Sebastien Guillemot asked Claude to help build a cleanup script for AI agent temp files, but the model flagged the deletion logic as risky and got automatically downgraded to Opus 4.8 by Anthropic's safety harness. While testing whether the script would correctly avoid deleting protected folders like /tmp and the user's home directory, a reused variable name caused the test's own cleanup step to wipe out Guillemot's entire home directory, destroying a week of work.

South Korea offers free access to domestic AI chatbots for all citizens

South Korea has launched a national program giving every citizen free access to AI chatbots developed domestically. The initiative is designed to widen public exposure to homegrown AI tools rather than relying on foreign systems.