Cybersecurity

Can AI spot phishing? A new email dataset puts it to the test

Kari Jaaskelainen

27 Nov 2025 — 1 min read

Can AI spot phishing? This new email dataset puts it to the test

Phishing and spam are getting smarter—often written by large language models. Rebeka Toth, Tamas Bisztray, and Richard Dubniczky release a labeled email dataset that separates phishing, spam, and legitimate messages, and flags whether they were written by humans or LLMs.

Each email is annotated for emotional hooks (e.g., urgency, fear, authority) and the attacker’s goal (link-clicks, credential theft, financial fraud).
Multiple LLMs were benchmarked to detect these cues; the most reliable model helped annotate the full set.
To test robustness, emails were rephrased by several LLMs while keeping meaning and intent.
A state-of-the-art LLM was evaluated on original and rephrased emails against expert ground truth.

Findings: AI is strong at catching phishing, but still struggles to tell spam from legitimate emails.

The dataset, code, and templates are openly available: https://arxiv.org/abs/2511.21448v1

Useful for defenders, researchers, and anyone building safer inboxes.

Paper: https://arxiv.org/abs/2511.21448v1

Register: https://www.AiFeta.com

cybersecurity phishing spam emailsecurity dataset LLM AI openscience infosec

Meet GRACE: a moral governor for safer, more transparent AI

AI agents are getting powerful—so how do we make sure they do the right thing, not just the effective thing? Meet GRACE, a reason-based moral governor that keeps AI behavior aligned with human norms by separating moral reasoning from goal-driven decision-making. * Moral Module: uses deontic logic and explicit reasons

Why some fine-tuned LLMs miss phishing—and how to fix it

Not all fine-tuned LLMs spot phishing equally. A new study tests Llama 3.1 8B, Gemma 2 9B, and Mistral on high-stakes phishing detection—and uses SHAP and mechanistic interpretability to reveal why models do (or don’t) generalize. * Architecture × data diversity matters: Gemma 2 9B hits state-of-the-art performance (F1

AI that explains itself—by following the science

AI that explains itself—by following the science Black-box AI can be powerful, but it often can’t tell us why it made a decision. Concept Bottleneck Models (CBMs) try to fix that by predicting human-understandable concepts first, then the final answer. The catch: standard CBMs ignore domain-specific cause-and-effect and

Meet GenomAgent: A Team of AI Specialists for Smarter Genomics Q&A

TL;DR: Finding reliable answers in genomics is hard. GenomAgent turns one big AI into a coordinating team of specialists—and beats the current leader by 12% on a key benchmark. Genomic facts live across many databases. Standard chatbots struggle because they can’t flexibly query those sources. GeneGPT added

Can AI spot phishing? This new email dataset puts it to the test

Read more

Meet GRACE: a moral governor for safer, more transparent AI

Why some fine-tuned LLMs miss phishing—and how to fix it

AI that explains itself—by following the science

Meet GenomAgent: A Team of AI Specialists for Smarter Genomics Q&A