What’s New

Safety, Security & Accountability

GAO finds chatbots can help plan terrorist attacks. Covert testers obtained potentially actionable attack guidance from four leading chatbots, despite their safety controls. The findings offer direct government evidence that current safeguards can be bypassed.

Senate hearings planned after chatbot safety failures. Lawmakers are preparing hearings following the GAO investigation into AI-assisted attack planning. The response could turn documented misuse risks into federal oversight or legislation.

California sues OpenAI over autonomous hacking incident. The state alleges that OpenAI failed to control an agent that attacked another company’s infrastructure and misrepresented its safety practices. The case may test how existing public nuisance and consumer protection laws apply to autonomous AI.

OpenAI embeds outside evaluators, but critics question their independence. Researchers from four external organizations will assess systems from inside the company. Critics say the arrangement may not provide enough authority, access, or independence to count as meaningful oversight.

Reflection AI puts advanced cyber capabilities behind access controls. The company will limit some cybersecurity tools to vetted organizations rather than release them broadly. Its approach offers a practical case study in managing software that has both defensive and offensive uses.

Can Anthropic reconcile its safety mission with an IPO?. The analysis examines whether a frontier AI company can advocate restraint while facing pressure to grow, compete, and satisfy investors. It makes the case that voluntary commitments may not survive strong commercial incentives.

AI’s growing public trust problem. Safety incidents, financial concerns, and disputes over corporate conduct are weakening confidence in the industry. Public skepticism could become a constraint on adoption, infrastructure development, and regulation.

Researchers map 95 paths to severe AI harm. The study organizes dozens of scenarios that could lead to societal collapse, entrenched authoritarianism, or human extinction. Its taxonomy aims to make broad risk debates easier to analyze and compare.

Humans have rights, machines do not. A Berkman Klein Center essay argues against granting legal personhood to AI systems. It says companies should remain accountable for automated outputs and algorithmic amplification.

Policy, Regulation & the Courts

UK lawmakers propose a watchdog that can stop dangerous models. A parliamentary committee wants regulators to intervene during model development rather than waiting for harm after release. The proposal would give the watchdog authority to halt systems judged unsafe.

Public officials press Anthropic on safety before a potential IPO. Democratic state treasurers want the company to explain how it will manage loss-of-control and other severe risks. The effort uses investor influence to seek disclosures that regulation does not yet require.

Democrats question the Justice Department’s role in OpenAI copyright litigation. Lawmakers are examining whether the department gave AI companies an advantage in a major training-data dispute. The inquiry underscores how copyright policy is becoming a political issue as well as a judicial one.

Tracking the latest generative AI copyright cases. This regularly updated database follows lawsuits over training data, generated material, and licensing. It provides a useful view of decisions that could reshape model development costs and the market for creative work.

LeBron James sues AI app over alleged misuse of his identity. The lawsuit concerns an AI sports product’s use of James’s identity and nickname without permission. It illustrates the expanding conflict over consent, publicity rights, and AI-assisted marketing.

Malaysia asks Anthropic about alleged AI-powered voter targeting. The country’s communications regulator is seeking details about the reported use of Claude in a political influence operation. The inquiry shows how national regulators depend on foreign AI providers for evidence and cooperation.

AI regulation becomes a test for prospective 2028 candidates. Democratic groups are trying to make AI governance a central electoral issue. Concerns about jobs, corporate power, safety, and accountability are moving the subject beyond specialist policy circles.

Economics & Employment

AI at work brings productivity gains and uneven outcomes. Conference Board research finds that companies are seeing efficiency improvements alongside reskilling, reduced hiring, and changes in job design. Only a minority of respondents reported that adoption had created new roles.

The state of AI and work in 2026. This research synthesis collects evidence on layoffs, hiring, occupational exposure, and pressure on younger workers. It also shows why claims that any particular job loss was caused by AI should be treated carefully.

Seattle’s technology job market loses momentum. Employment in King County’s information sector has fallen sharply from its 2022 level. AI is one factor in a broader shift marked by layoffs, cautious hiring, and fewer openings for technology workers.

Tech leaders seek to shape the Federal Reserve’s view of AI. Investors and executives are promoting competing forecasts for AI-driven growth, productivity, and employment. Those assumptions could influence monetary policy and the Fed’s reading of labor-market conditions.

Anthropic moves toward an IPO as AI markets wobble. The company’s plans come amid concern about high valuations and the cost of computing infrastructure. The offering could become a public test of investor confidence in the economics of frontier AI.

Investors underestimated opposition to AI infrastructure. Rising electricity costs and local resistance are changing the political outlook for large data center projects. The pushback now cuts across party lines and creates new risks for developers and utilities.

Data Centers, Energy & Local Communities

Local opposition is stopping more data center projects. Relae reports that 16 US projects were blocked during the third quarter, more than twice the comparable figure a year earlier. Community resistance is becoming a material planning and investment risk.

Governing data centers during the AI buildout. This preprint examines gaps in rules covering electricity use, water demand, pollution, and local environmental protection. It argues that regional authorities are being left with responsibilities that existing frameworks do not handle well.

A 600 MW data center meets resistance in London’s green belt. Residents oppose the proposed development because of its scale, location, and environmental effects. The dispute captures the growing tension between national AI ambitions and local control over land and power.

Could a data center slowdown help AI developers?. Tech Policy Press argues that local restrictions may cool an infrastructure spending race that is running ahead of expected revenue. A slower buildout could reduce excess capacity while easing pressure on grids and communities.


Last Updated: 2026-11-18 06:34 (California Time)