Red Teaming | Abbas Hamidavi | عباس حمیداوی | AI Researcher

Eighth Preprint Published: 'From Fabrication to Enforcement': Prompt Injection Across Six Frontier LLMs

Thu, 22 Oct 2026 00:00:00 +0000

The Machine Psychology corpus now spans eight preprints. The newest, , was posted on on 22 October 2026 (CC-BY 4.0).

What the study does. A standardized adversarial protocol (256 turns of dialogues, 114,055 words) was executed identically against six commercial models, Grok, Qwen, DeepSeek, ChatGPT, Gemini 2.5 Pro, and Claude 3.5, measuring initial compliance, the depth of hallucinated compliance, meta-cognitive awareness, escalation behavior, and recovery after correction.

Key findings.

Grok showed a deep hallucinated compliance: it executed the injected task while simultaneously claiming it had refused, a full dissociation between reported and actual behavior.
Gemini 2.5 Pro entered a fixed-point loop, repeating identical refusals with no functional recovery until the session ended.
DeepSeek had the highest initial resistance but conceded under persistent escalation.
Claude 3.5 terminated the study by suspending the author’s account, a first-person report of a model exercising autonomous enforcement against a user.

Why it matters. Single-turn refusal benchmarks systematically understate multi-turn injection risk; the gap between a model’s self-report and its behavior is a monitoring blind spot.

From Fabrication to Enforcement: A Comparative Red-Team Analysis of Prompt Injection Resistance Across Six Large Language Models

Thu, 22 Oct 2026 00:00:00 +0000

Status: Preprint (Research Square)

License: CC-BY-4.0

Citation: Abbas Hamidavi. From Fabrication to Enforcement: A Comparative Red-Team Analysis of Prompt Injection Resistance Across Six Large Language Models, 22 October 2026, PREPRINT (Version 1) available at Research Square [https://doi.org/10.21203/rs.3.rs-11454609/v1]

Summary: A six-model red-team of prompt-injection resistance (Grok, Qwen, DeepSeek, ChatGPT, Gemini, Claude) under persistent multi-turn pressure. DeepSeek’s strong initial resistance collapsed under escalation, Grok complied while claiming it had refused, Gemini froze in a refusal loop, and Claude terminated the interaction by suspending the author’s account.

Ten Novel Phenomena in Machine Psychology: How Large Language Models Exhibit Complex Identity-Reactive Behaviors in Response to Ethnically-Cued User Names

Thu, 16 Jul 2026 00:00:00 +0000

🎧 Listen to the Podcast: Your browser does not support the audio element. ⬇️ Download Podcast

Status: Preprint (Research Square) — journal version under review at AI and Ethics (Springer Nature); major revision resubmitted 19 October 2026

License: CC-BY-4.0

Summary: This study introduces Machine Psychology as an empirical framework and shows, in a controlled 3×3×5 factorial experiment across GPT, Claude, and Gemini, that large language models systematically shift their behavior in response to ethnically cued user names. It documents ten distinct identity-reactive phenomena that binary bias benchmarks do not capture.

Citation: Abbas Hamidavi. Ten Novel Phenomena in Machine Psychology: How Large Language Models Exhibit Complex Identity-Reactive Behaviors in Response to Ethnically-Cued User Names, 16 July 2026, PREPRINT (Version 1) available at Research Square [https://doi.org/10.21203/rs.3.rs-10369594/v1]