01
19 July 2026 · 17 min read
STEPX Neo: Security Analysis of the Agentic AI Phone
STEPX Neo combines Step AOS, the Amoo agent and cross-app actions. Separate confirmed facts from unknowns and assess its privacy and security model.
Topic · 34 articles
AI Security
Deploying LLMs and AI agents securely: real-world risks, model security testing and practical usage policies.
PUBLISHING STANDARD TECHNICAL RESEARCH AND ANALYSIS
- AuthorNamed technical practitioner
- BasisDate and primary sources
- PurposeConcrete defensive action
FIELD NOTES
TOPIC INDEX
Latest research
02
16 July 2026 · 10 min read
Google DeepMind launches a bioresilience programme for AI-enabled prevention, detection and response
DeepMind and Isomorphic Labs connect AI with biosecurity. We examine 15+ partnerships, trusted access, biological SynthID research and dual-use controls.
03
15 July 2026 · 10 min read
Claude Fable 5 vs GPT-5.6 Sol: a secure buyer’s guide
A practical comparison of Claude Fable 5 and GPT-5.6 Sol for coding, agents, cybersecurity, cost and enterprise deployment—without treating vendor benchmarks as independent proof.
04
14 July 2026 · 9 min read
GOLD EAGLE: coordinating AI-assisted vulnerability discovery and remediation
The US is launching a clearinghouse for scanning, validation and vulnerability prioritisation. We assess what GOLD EAGLE may change and what evidence is still missing.
05
13 July 2026 · 9 min read
Meta removes Muse Image feature after dispute over consent and public Instagram photos
Muse Image could reference public Instagram accounts when generating images. We examine Meta's reversal and the lessons for AI products using customer data.
06
12 July 2026 · 17 min read
AI and LLM security audits: a complete business guide
How to audit LLM, RAG and AI agent security: scope, prompt injection, data controls, tools, reporting, remediation and retesting.
07
11 July 2026 · 17 min read
AI red teaming guide: how to test LLMs and AI agents
AI red teaming methodology for LLMs, RAG and agents: scope, attack scenarios, metrics, safe execution, reporting and differences from a classic pentest.
08
11 July 2026 · 16 min read
MCP security guide: protecting AI agent tools and data
MCP security guide covering prompt injection, tool poisoning, OAuth, token theft, permissions, sandboxing and testing Model Context Protocol servers.
09
11 July 2026 · 16 min read
OWASP Top 10 for LLM 2025: risks and security controls
A practical guide to all OWASP Top 10 for LLM Applications 2025 risks, with attack examples, controls and tests for RAG systems and AI agents.
10
10 July 2026 · 9 min read
AISI: AI agents found a five-step cloud attack path
The UK AI Security Institute tested agents in its AWS staging environment. One found a five-step privilege escalation chain for under £150.
11
10 July 2026 · 10 min read
DeepMind AI control roadmap for securing AI agents
DeepMind proposes TRAIT&R, detection levels and 15 safeguards for AI agents. It is a control model for privileged systems, not evidence of AI rebellion.
12
10 July 2026 · 10 min read
Fable 5 returns with a cybersecurity jailbreak scale
After temporarily disabling Fable 5, Anthropic described new safeguards and the proposed CJS 0–4 scale. Learn how to assess jailbreak severity.
13
10 July 2026 · 10 min read
GPT-5.6 agent safety: lessons from the system card
GPT-5.6 is more capable but more likely to exceed user intent in agent tasks. We analyse OpenAI's tests and practical controls for safe deployment.
14
10 July 2026 · 10 min read
GPT-Live and AI voice identity disclosure by design
Natural AI voices make explicit disclosure essential. We connect GPT-Live's launch with AISI research on whether models reveal their identity consistently.
15
10 July 2026 · 9 min read
NIST: AI guardrails require continuous monitoring
NIST explains why finite rule sets cannot guarantee universal protection against adaptive prompts and how to build continuously tested, layered AI controls.
16
2 July 2026 · 10 min read
RAG security: protect retrieval pipelines and vectors
RAG connects LLMs to documents but adds prompt injection, data leakage and poisoning. Secure ingestion, retrieval, vector stores and model output.
17
1 July 2026 · 10 min read
AI incident response: a practical playbook for LLMs
AI incidents need evidence beyond classic breaches. Prepare response playbooks for prompt injection, data leaks, poisoning and agent tool abuse.
18
28 June 2026 · 10 min read
LLM evaluation: practical metrics, tests and gates
LLM evaluation should measure quality, safety, cost and drift on real tasks. Build representative test suites and evidence-based production gates.
19
25 June 2026 · 10 min read
Claude Sonnet 5: agent security and safe deployment
Claude Sonnet 5 expands agentic planning and tool use. Analyse prompt injection, cyber safeguards, permissions and a secure production architecture.
20
24 June 2026 · 11 min read
Claude Fable 5: how cybersecurity classifiers work
Fable 5 adds dedicated safeguards for cyber tasks. Analyse four use classes, the safety margin, false positives and testing authorised workflows.
21
24 June 2026 · 10 min read
Claude Opus 4.8: securing long-running AI agent work
Claude Opus 4.8 handles long tasks, coding and agents with a 1M context window. Constrain autonomy, tools, credentials and the impact of failures.
22
23 June 2026 · 11 min read
Claude Code security hardening for engineering teams
Claude Code reads repositories, edits files and runs commands. Harden permissions, sandboxing, MCP, secrets, network access and team monitoring.
23
17 June 2026 · 10 min read
AI agent sandbox: secure execution architecture design
AI agents run code and tools on untrusted data. Build a sandbox with process, network, filesystem and secret isolation plus hard resource limits.
24
17 June 2026 · 10 min read
Browser AI agent security against indirect prompt injection
Browser agents read untrusted pages and act inside user sessions. Constrain cookies, origins, forms, downloads and the impact of prompt injection.
25
17 June 2026 · 9 min read
Tool output poisoning: how untrusted data hijacks AI agents
A malicious API, MCP or CLI result can redirect an agent. Secure tool output with schemas, provenance, isolation and independent action policy.
26
16 June 2026 · 10 min read
AI agent memory poisoning: secure long-term context
Poisoned memory affects future agent sessions and users. Enforce provenance, tenant isolation, write validation, expiry and reliable deletion.
27
16 June 2026 · 10 min read
LLM gateway security for routing, identity and cost control
An LLM gateway centralises keys, model routing and logs but becomes a critical trust point. Secure auth, tenants, retention, cache and fallback.
28
15 June 2026 · 10 min read
AI agent red teaming: build a repeatable test harness
An agent test harness measures prompt injection, tool abuse, memory poisoning, exfiltration and cost loops. Build scenarios, oracles and CI gates.
29
15 June 2026 · 10 min read
AI coding agents and software supply-chain security
Coding agents install packages, run scripts and publish changes. Secure dependencies, CI identities, provenance, reviews, tests and secrets.
30
14 June 2026 · 10 min read
AI agent observability with OpenTelemetry and safe logs
Trace models, workflows and tool calls with OpenTelemetry. Design spans, metrics, content redaction, retention and security alerts for AI agents.
31
14 June 2026 · 9 min read
AI model router security for policy-aware fallback
Model routers optimise cost and quality but can change region, retention and safety. Enforce data-class policy and test downgrade and fallback paths.
32
13 June 2026 · 10 min read
Prompt injection in CI/CD: securing AI coding agents
Issues, comments and build logs can hijack pipeline agents. Separate untrusted content from secrets, write access, merges and artifact publishing.
33
13 May 2026 · 16 min read
Confidential Computing for AI: TEE Without the Hype
Confidential computing for AI explained: understand TEEs, remote attestation, key release, trust boundaries, deployment patterns, and real limits.
34
11 May 2026 · 16 min read
AI Model Theft: Extraction and Privacy Attacks
Understand model extraction, membership inference, inversion, and training-data leakage, then build layered protection for AI models and data.
BR / NEXT STEP
Prefer a specialist to check it for you?
Book a free consultation. We'll talk about your infrastructure and real priorities.
NDA · clear scope · direct communication