AI-Written Emails Are Getting Caught as Spam More Often Than Human Ones
New research from Folderly finds AI-written cold emails hit spam filters more often and get lower open and reply rates than human copy.
Anthropic Deploys Lobbyists in Washington as Export Controls Threaten Global Claude Rollout
By Prompt AI News2 min read
#anthropic#policy#export-controls#lobbying
The Wall Street Journal reports that Anthropic dispatched a team to Washington this week to combat new AI export restrictions threatening to cap where the company can deploy Claude internationally. The lobbying effort signals the restrictions are serious enough for one of the best-funded AI labs in the world to treat them as an emergency.
The controls, part of broader U.S. government efforts to limit the spread of advanced AI to adversarial nations, would constrain Anthropic's ability to sell or license Claude in restricted markets. For a company that has raised over $7 billion and burns through capital at a startup's pace, losing access to international customers is a direct threat to its path to profitability.
The timing is awkward. Stratechery's Ben Thompson published a piece this week calling AI safety Anthropic's competitive superpower, arguing the company's principled stance distinguishes it from OpenAI. But Anthropic's most pressing challenge right now isn't convincing the public it's the responsible choice — it's convincing the U.S. government to let it do business abroad.
The AI industry's Washington moment has arrived. Expect more companies to staff up inside the Beltway as regulators increasingly treat frontier models not as software products but as strategic national assets.
Leave a Comment
All comments are reviewed before appearing. Keep it respectful.
More in Commentary
View all →Conservative Coalition Targets AI Data Centers in Nationwide Protest Campaign
Right-leaning rural groups are organizing multi-state protests against AI data center expansion, citing power draw, water use, and local grid strain.
Open-Source Tool Queries 11 AI Models at Once to Filter Hallucinations
A developer's open-source Quorum system cross-checks 11 LLMs simultaneously, surfacing answers only when a supermajority agrees — trading cost for reliability.