Comment by gruntled-worker
5 hours ago
There's something missing from the picture. The bots are:
- Using a terribly inefficient way to redownload the same commits as e.g. HTML diffs, possibly the most inefficient.
- Putting in tons of CPU cycles to surpass the Anubis PoC.
- Putting in other kinds of active effort like reworking access methods and buying "residential proxies" that are probably illegal in most jurisdictions.
This sounds more like escalating DDoS than AI scraping.
Possible, but I doubt it. The sheer number of other free-to-read content sites dealing with the exact same problem the last few years (many of whom there's no plausible reason to DDoS) tells us that this is content harvesting, not an attack.
I'm curious and would like to see those reports. There's AI scraping for sure, but intentionally resource-consuming, increasingly-insidious AI scraping I've never actually read about.
AI scraping might be bad, but if a particular case that's actually a DDoS becomes the cause celebre against AI scraping, it will weaken the argument, not strengthen it.