FTC deletes Lina Khan-era AI posts on risks and open weights

The Federal Trade Commission has deleted several Lina Khan–era blog posts warning about artificial intelligence risks and debating the role of open-source and open-weights models. The removal followed the administration change, as reported by Wired. According to the report, the takedowns repositioned the agency’s behavior while recasting how Washington talks about AI, competition, and consumer protection.

Among the entries deleted were the posts titled:

• "On Open-Weights Foundation Models"
• "Consumers Are Voicing Concerns About AI"
• "AI and the Risk of Consumer Harm"

They laid out the FTC’s examination of AI’s real-world harms — fraud, impersonation, discriminatory outcomes, and the explosion of commercial surveillance — while exploring whether looser model release practices could make such outcomes more common. The open-weights post focused on the industry’s blurry line between open source and permissive access to model weights. That blur mattered. As O’Reilly explained, the Open Source Initiative and prominent technical groups have insisted that label choices determine who can audit systems, reproduce research, and mount a cogent critique. The FTC’s now-deleted analysis placed those questions in a consumer protection frame: who pays the price when the powerful weapon is widely available and poorly secured.

Policy shifts reflect a faster, more open federal AI posture

The removals reflect a broader federal climate that prioritizes accelerated AI deployment and open-source movements that reduce duplication and allow for quicker, more competitive global action. Wired recently reported that the FTC deleted hundreds of additional items linked to AI, consumer protection, and agency lawsuits brought against Big Tech platforms, signaling that leaders are no longer governed by caution.

Today, any firewall between proponents of the new kind of federalism — claiming openness advances security via transparency, startup generation, and strengthening corporate power — versus how much “open” is good enough gets rid of essential guardrails and broader CSL priorities. Critics of the new trend say that “open” circulation can decrease the threat level, primarily as model capabilities evolve, with the removal of the FTC posts trying to navigate that ever-tightening needle.

Records laws and transparency concerns after deletions

The elimination of policy analyses raises tricky frontier-related records questions. The Federal Records Act mandates that documentation for numerous crucial organizational functions and decisions must be deposited. The Open Government Data Act stipulates that “agencies should prioritize performance,” and more specifically that “agencies will foster public access and full use of unrestrained and open public data, and keep open data inventories in two main places.”

Good-government organizations like the Electronic Privacy Information Center and the Center for Democracy and Technology have long pushed for the AI-oriented recommendations to be left untouched because they influence enforcement and commercial activity. The prior FTC bosses frequently added alerts or contextual information to the legacy content instead of deleting them completely, in a series crafted to maintain the activity without committing to every prior move.

Enforcement signals amid a shifting global AI landscape

The takedowns remove public breadcrumbs about how the FTC is evaluating AI risks in fraud, privacy, and discrimination cases. This is a period when the market starts consolidating via acquisitions and acqui-hires. Even when not binding, staff blogs and Office of Technology posts qualify as early indicators for enforcement priorities. The signals impact companies’ decisions to delay product launches and updates as well as disclosures.

The broader context is an ongoing high-stakes regulatory race. The European Union has enacted the AI Act, setting obligations by risk tier. The United Kingdom opted for the sector-led approach, with the guarantee of model evaluations by the AI Safety Institute. The United States went for a more decentralized calibration. It makes the FTC’s thought leadership more critical for guiding norms around deceptive claims, unfair practices, and data misuse in AI.

Silence on the tech blog and signals from FTC updates

According to Wired, the number of posts added by the FTC’s Office of Technology Blog remains the same. No fresh entries have been added by the administration. The observation is notable due to the fast pace of frontier-model releases and the spread of genAI features in consumer apps. All these efforts should be supplemented, given the still numerous allegations linked to platforms’ power. Without updated guidance, the industry will have to rely on complaints, consent orders, and speeches to guess the direction.

What to watch next: oversight, disclosures, and FOIA

The next alternative is congressional oversight requests, disclosures, and Freedom of Information Act inquiries that would look for the inventories. The focus is on the number of removed posts and the archival status. If the posts were scheduled and preserved, the question will shift to policy goals instead of the compliance issue.

Implications for AI builders and everyday consumers

For AI builders, the upshot is the same as before: safety, accuracy, and bias allegations are still marketing and product claims and hence still fall under falsity legislation and unfairness criteria. For consumers, less public exposure by presiding institutions means that safeguards rely on independent monitoring bodies, community groups, and the press to highlight risks. Additionally, how it is conveyed — not just policed — will define the equilibrium between velocity and security in the future stage of AI.