Germany's far-right AfD: half a million internal documents leaked
A leak of roughly half a million internal emails from Germany’s far-right AfD party has raised questions about both the authenticity of the data and the way it is being presented via AI-generated summaries on hexane.cc. Commenters debate whether it is reasonable to rely on those summaries without manually inspecting the 300GB torrent of raw messages, how DKIM and media reports (e.g. Reuters, DW) support verification, and why Cloudflare and some platforms have flagged or downranked the site. The exchange also broadens into concerns about political bias, free-speech and defamation standards in Germany, and whether targeted leaks of a single party unfairly shape public perception.
Leak access, authenticity, and verification
- Corpus is distributed via large torrents (~300 GB), with both packed and per-file variants.
- Some participants downloaded samples and report that emails match the website’s AI-generated summaries.
- A “99_FOR_JOURNALISTS/VERIFY.md” file is said to describe DKIM-based verification; several commenters say messages can be cryptographically validated.
- External German and international media are cited as having treated the leak as at least partly genuine.
- Others remain wary, arguing the burden of proof lies with the site’s authors, not with readers downloading huge datasets.
AI-generated summaries and “AI slop” concerns
- The site relies heavily on AI summaries of leaked emails, often without showing the underlying text.
- Critics argue this is “AI slop”: ungrounded, incompletely documented, and vulnerable to hallucinations.
- Requests are made for clear methodology, prompts, models used, and side-by-side samples to check fidelity.
- Supporters counter that the full corpus is available for anyone to audit, and some see this as an “excellent use of AI” for triaging large archives.
Usability, scale, and user expectations
- Some find downloading 300 GB or navigating torrents an unreasonable expectation just to validate claims.
- Others argue this is trivial by modern standards and that complaining instead of sampling the data is “intellectual laziness.”
Cloudflare phishing warning and site trust
- Cloudflare displays a “suspected phishing” warning for the site, which some interpret as evidence of active attempts to discredit it, others as potential shadiness.
- Monero donation solicitations add to perceived ambiguity.
German speech and defamation context
- Debate over whether German law (defamation, “Volksverhetzung,” insult offenses) makes speech significantly less free than in the US.
- One side stresses Germany lacks US-style absolute speech protection; the other notes Germany’s constitutional free-expression guarantees with different limits.
HN flagging and perceived political bias
- The story was flagged; reasons suggested include:
- General aversion to political content on HN.
- Alleged tech-community sympathy for far-right movements and desire to suppress damaging stories.
- Others dispute systemic bias, suggesting politics simply doesn’t fit HN’s core focus.