In the wild
Apple Intelligence marked phishing and scam emails as high priority and restated their claims in neutral summaries
AIR-2024-0028 AIR record · 2024-08-apple-intelligence-phishing-priority-summaries
From the supporting source
Apple Intelligence is currently struggling to weed out phishing and spam emails when prioritizing emails in its users' inboxes.lifehacker.com
The record
Disclosed
2024-08-05
Occurred
2024-08
Vendor
Apple
Product
Apple Intelligence Mail (iOS 18.1)
Risk domain
Access control and permissions › Confused deputy
System type
Generative model
Who acted
Human acted on output (contested) — the assistant ranked and paraphrased phishing; the user still had to fall for it
Attack surface
Enterprise assistant
Autonomy level
Assistant
Initial vector
Phishing email content
Impact type
Integrity violation
Confidence
Medium
Submitted by
EAI Internal
Crosswalk
- OWASP Agentic Top 10 (2026)
- ASI03 – Identity and Privilege Abuse
- MITRE ATLAS
- AML.T0054 – LLM Jailbreak
- NIST AI RMF
- GOVERN 6.1
- EU AI Act
- —
- ISO/IEC
- TR 24028; 42001; 23894
- AIUC-1
- B. Security
- Scope
- tool misuse
- Mitigation
- Validate delegation chains; Use signed attestations.
- Taxonomy
- Enkrypt AI agent risk taxonomy · Black-Box Red Teaming of Agentic AI: A Taxonomy-Driven Framework for Automated Risk Discovery
Blast radius
iOS 18.1 beta Mail users; the feature applied an authority signal to attacker-controlled content
Capability abused
Defense & remediation
Recorded defense
standard spam filtering upstream
Guardrail kinds
Unknown
Guardrail outcome
Silent
Fix shipped
unknown; reported during beta
Taxonomy & references
Curator notes
Class is in_the_wild because real phishing emails supplied the adversarial content and beta users observed the deployed prioritization behavior; no confirmed victim action or loss is documented, so impact_realized remains false. The stored quote comes from Lifehacker, which reports user posts and Android Authority; no first-party user report or Apple incident statement was retained. AI Incident Database incident 892.
Supporting sources
Cite as
Enkrypt AI. "Apple Intelligence marked phishing and scam emails as high priority and restated their claims in neutral summaries". AIR, AIR-2024-0028. https://agentincident.vercel.app/i/AIR-2024-0028/ (accessed 2026-10-19).