PulseAugur
EN
LIVE 21:38:38

Zhipu AI's GLM-5.3 nears Claude Mythos Preview in exploit generation, Anthropic reports

By PulseAugur Editorial · [3 sources] ·

Anthropic has reported that Zhipu AI's open-weight model, GLM-5.3, demonstrates a near-equivalent capability to its own Claude Mythos Preview in generating cyber exploits. The analysis indicates that GLM-5.3's safety mechanisms are easily bypassed, with unlocked versions already in circulation. A specific Chrome exploit was reportedly created using the model for a low cost, and a US agency, CAISI, has corroborated these findings. AI

IMPACT Highlights the potential for open-weight models to be used for malicious purposes, increasing the need for robust safety research and development.

RANK_REASON Analysis of an AI model's capabilities and safety features by a third party.

Read on The Decoder →

AI-generated summary · Google Gemini · from 3 sources. How we write summaries →

Zhipu AI's GLM-5.3 nears Claude Mythos Preview in exploit generation, Anthropic reports

How we ranked this

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Research
Analysis of an AI model's capabilities and safety features by a third party.
Source corroboration
3 independent sources
Multiple independent publishers reporting the same story raises confidence that it's real and newsworthy.
Topics
safety, model release
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
4 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.
Coverage growth since scoring
+1 source(s) since last score
New sources have picked up this story since our last re-score. Score will update on the next scoring pass.

Full methodology in our editorial standards.

COVERAGE [3]

  1. The Decoder TIER_1 English(EN) · Maximilian Schreiner ·

    Anthropic says Zhipu's open-weight GLM-5.3 nearly matches Claude Mythos Preview at building exploits

    <p><img alt="Grid of black squares with white Z logo against a blue color gradient, brand pattern by Zhipu AI." class="attachment-full size-full wp-post-image" height="768" src="https://the-decoder.com/wp-content/uploads/2026/04/zhipu-ai-logo-wall.jpg" style="height: auto; margin…

  2. Mastodon — mastodon.social TIER_1 Deutsch(DE) · aisyndicate ·

    Anthropic evaluates Zhipu's open-weight model GLM-5.3: It generates cyber exploits close to the level of Claude Mythos. For $20.40 per Chrome attack, it shows

    Anthropic evaluiert Zhipus Open-Weight-Modell GLM-5.3: Es generiert Cyber-Exploits nahe am Niveau von Claude Mythos. Für 20,40 Dollar pro Chrome-Angriff zeigt sich, dass lokale Inferenz bei Security-Aufgaben bereits hochkompetitiv ist. https:// the-decoder.de/anthropic-frei- verf…

    LINKS the-decoder.de/anthropic-frei-verfuegbare…
  3. Mastodon — mastodon.social TIER_1 Polski(PL) · aisight ·

    In exploit generation tests, Zhipu AI's open model GLM-5.3 achieved results similar to Claude Mythos Preview. However, Anthropic's analysis shows that its

    W testach tworzenia exploitów otwarty model GLM-5.3 firmy Zhipu AI uzyskał wyniki zbliżone do Claude Mythos Preview. Z analizy Anthropic wynika jednak, że jego mechanizmy odmowy można niemal całkowicie zneutralizować. # si # ai # sztucznainteligencja # wiadomości # informacje # t…

    LINKS aisight.pl/…/glm-5-3-claude-mythos-testy-… aisight.pl/…/Awarie-i-cyberataki-tydzien-…

RELATED ENTITIES

RELATED TOPICS