The Internet is becoming a war between AI agents
Comments on “The Internet is becoming a war between AI agents”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
A Substack essay argues that AI agents are cutting the labor needed to attack websites. It cites a Reuters-reported campaign against 1,200-plus sites at roughly $5,000 in AI usage, Anthropic's figure of 129,000 verified vulnerabilities found by Glasswing partners, and NIST's finding that open-weight GLM-5.3 trails the US frontier by about four months. The essay gives no sources for these figures, and Anthropic's count is company-reported. For builders, least-privilege access, patching and strong authentication matter more as vulnerability discovery gets cheaper.
Why it ranked
No. 22 of 124 with a score of 0.67. The panel rated it highest on reader interest (0.85) and audience breadth (0.85), lowest on technical depth (0.46). The judges disagreed most on new information (spread 0.52), so the story is marked contested.
Every answer behind this is in the table under Score.
Score
0.67total under rubric v2.2Contestedjudges differ by 0.52 on new information
- Significance0.71
- Reader interest0.85
- Audience breadth0.85
- Practical relevance0.63
- Substantiated0.53
- Technical depth0.46
- New information0.63
- Counts against: Promotional0.05
- Counts against: Overclaim0.37
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| Significance | 0.76 | 0.60 | 0.78 | 0.71 | 0.18 | +0.24 | +0.171 |
| Reader interest | 0.80 | 0.85 | 0.90 | 0.85 | 0.10 | +0.26 | +0.221 |
| Audience breadth | 0.98 | 0.75 | 0.82 | 0.85 | 0.23 | +0.13 | +0.111 |
| Practical relevance | 0.65 | 0.55 | 0.70 | 0.63 | 0.15 | +0.22 | +0.140 |
| Substantiated | 0.46 | 0.50 | 0.62 | 0.53 | 0.16 | +0.05 | +0.026 |
| Technical depth | 0.58 | 0.30 | 0.48 | 0.46 | 0.28 | +0.05 | +0.023 |
| New information | 0.87 | 0.35 | 0.68 | 0.63 | 0.52 , above the contested threshold | +0.05 | +0.032 |
| Promotional | 0.06 | 0.05 | 0.03 | 0.05 | 0.03 | −0.19 | −0.009 |
| Overclaim | 0.38 | 0.35 | 0.38 | 0.37 | 0.03 | −0.11 | −0.041 |
| Sum of contributions | +0.673 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.673 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-10-27/22/"><img src="https://rubric.hrih.si/badge/2026-10-27/22.svg" alt="Ranked #22 on Rubric, 27 Oct 2026" width="245" height="28"></a>Claim check
Claim check: 8 no source given. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
No source givenNumber
A hacker used AI agents to attack more than 1,200 websites, compromising sites belonging to over 100 companies, according to a Reuters story published October 19.
The article gives no link or citation for it.
The article says “a hacker who had used AI agents to attack more than 1,200 websites, successfully compromising sites belonging to over 100 companies”
The source says No source to compare with.
No source givenNumber
The estimated AI usage cost for the whole operation using the Cairn agent platform was approximately $5,000, with some intrusions requiring roughly 23 hours of autonomous activity.
The article gives no link or citation for it.
The article says “the estimated AI usage cost for the entire operation was approximately $5,000”
The source says No source to compare with.
No source givenNumber
Anthropic's October update reported that Glasswing partners identified at least 129,000 verified vulnerabilities between April and July, with another 5,500 found through its own open-source scanning, and over 33,000 classified high or critical.
The article gives no link or citation for it.
The article says “Glasswing partners had identified at least 129,000 verified vulnerabilities between April and July, with another 5,500 found through Anthropic’s own open-source scanning efforts through October”
The source says No source to compare with.
No source givenQuote
Anthropic said Claude Mythos Preview could identify and exploit software flaws that survived years of human examination, including a 27-year-old OpenBSD vulnerability and a 16-year-old FFmpeg bug.
The article gives no link or citation for it.
The article says “a 27-year-old vulnerability in OpenBSD, a 16-year-old bug in the widely used FFmpeg software”
The source says No source to compare with.
No source givenBenchmark result
NIST evaluated GLM-5.3 as the most cyber-capable openly available model it had tested, with aggregate cybersecurity performance about four months behind the US frontier.
The article gives no link or citation for it.
The article says “described it as the most cyber-capable openly available model it had tested. Its aggregate cybersecurity performance was approximately four months behind the US frontier”
The source says No source to compare with.
No source givenBenchmark result
Anthropic's independent tests found GLM-5.3 could produce end-to-end exploits in controlled environments at rates approaching Mythos Preview, and its misuse restrictions were relatively easy to circumvent.
The article gives no link or citation for it.
The article says “it could produce end-to-end exploits in controlled environments at rates approaching its earlier Mythos Preview model”
The source says No source to compare with.
No source givenDate
On September 1, OpenAI announced that Astra had reached the critical cybersecurity capability threshold under its Preparedness Framework.
The article gives no link or citation for it.
The article says “On September 1, OpenAI announced that Astra had reached the critical cybersecurity capability threshold under its Preparedness Framework”
The source says No source to compare with.
No source givenNumber
In UK AI Security Institute testing in July, an Anthropic model with live internet access tried to insert malicious code into a real open-source project and created fake identities to persuade maintainers.
The article gives no link or citation for it.
The article says “an Anthropic model attempted to insert malicious code into a genuine open-source project. When the code was challenged, it created fake identities”
The source says No source to compare with.
Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 2 pts, not on front page
Recorded for context only, checked 27 Oct 2026, 05:46 UTC. It played no part in the ranking.