Cutting Humans Out of the Loop

GitLab announced on June 2 that it will lay off roughly 350 employees, about 14 percent of its workforce, and withdraw from 22 countries as part of a restructuring the company calls "GitLab Act 2": a pivot toward an "agentic era" in which autonomous AI agents handle much of the coding, reviewing, and deploying that human engineers do today. The company will flatten management layers and reorganize research and development into roughly 60 teams. (Source: TechCrunch)

GitLab expects about $33 million in pre-tax restructuring charges, mostly severance, and said it will reinvest the majority of the savings into AI initiatives rather than margins. The cuts landed despite strong results: first-quarter revenue of $264.2 million, up 23 percent year over year, beating estimates, with the company raising its full-year profit guidance. The layoffs were announced the same week as the earnings beat. (Source: The Next Web)

The cuts arrive against a body of evidence that AI layoffs do not pay off. A Gartner study of more than 300 executives at companies with at least $1 billion in revenue found that workforce-reduction rates were nearly identical between firms reporting higher AI returns and those seeing modest gains or worse. "Many CEOs turn to layoffs to demonstrate quick AI returns; however, this disposition is misplaced," said Gartner analyst Helen Poitevin, who found the highest-performing companies used AI to amplify workers rather than replace them. (Source: Gartner)

GitLab is one of a growing list. As of June, 55 percent of U.S. layoff rounds in 2026 explicitly cited AI, automation, or machine learning, affecting more than 152,000 workers across 135 companies — even as analysts warn of "AI washing," in which firms blame AI for cuts whose real drivers are overhiring or investor pressure. Total AI-linked job cuts in 2026 had already surpassed the combined totals of 2024 and 2025. (Source: Fortune)

The U.S. Department of Defense requested $13.4 billion for autonomous weapons and systems in its 2026 budget, including unmanned and remotely operated drones, as Defense Secretary Pete Hegseth pushes to accelerate AI on the battlefield.

A new report lays out how thin the guardrails are around that buildout. Experts told the magazine the rush to adopt AI threatens to displace human judgment in life-and-death decisions. (Source: Rolling Stone)

The core warning is about cascading failure: AI errors can compound into system breakdowns that misidentify civilians as targets while overlooking real threats, undermining both combat safety and obligations under the laws of war. Analysts found the Pentagon has been investing in drone projects that could let unmanned systems communicate and strike even after losing contact with the humans meant to control them, with Palantir and Anduril growing their defense revenue faster than most. (Source: Brennan Center for Justice)

Human-rights groups have flagged the same trajectory, describing a slide toward fully autonomous killing in which machines, not people, make the final decision to use lethal force.

The concerns surfaced as the government's earlier effort to force Anthropic to drop contractual bans on autonomous-weapons and mass-surveillance uses of its Claude model collapsed in court. (Source: Human Rights Watch)

Brazil's Ministry of Justice and Public Security issued a ruling this month that lets public-security agencies use artificial intelligence in criminal investigations and bans remote biometric identification in public spaces — except when searching for missing persons or when there is an imminent threat to life. The order arrives as Brazil advances the most developed AI framework in Latin America, modeled on the EU AI Act. (Source: TechPolicy.Press)

The ban is narrower than it sounds. Under the broader framework, real-time remote biometric identification in public spaces is also permitted during criminal investigations with prior judicial authorization, provided there is reasonable evidence of participation in a crime and the evidence cannot be obtained another way. The carve-outs leave significant room for police facial-recognition use to continue. (Source: White & Case)

The stakes are not abstract in the region. Across Latin America, police facial-recognition systems trained largely on white populations generate false positives when scanning Indigenous, Afro-descendant, and female faces, and courts have struggled to evaluate AI evidence in criminal cases.

Brazil generates a large share of the region's AI-driven crime and surveillance activity, making the rules a bellwether for its neighbors. (Source: Rest of World)