So, the smart contract audit benchmarks tooling is completely finished, from dataset generation to reporting. The only thing that's missing is creating comprehensive baselines using the best available models. If anyone wants to sponsor this or has infinite API credits available lmk 🙂

https://github.com/muellerberndt/scabench

Attention, please

https://www.aikido.dev/blog/npm-debug-and-chalk-packages-compromised

It appears the Solana project 'Aqua' has likely rug pulled 21.77K SOL ($4.65M) after being promoted by teams such as Meteora, Quill Audits, Helius, SYMMIO, Dialect, and many influencers.

A few hours ago the funds were split four ways and transferred between intermediary addresses before being sent to multiple instant exchanges.

The team has since turned off replies on X (Twitter) for all posts.

Presale address
4Ea23VxEGAgfbtauQZz11aKNtzHJwb84ppsg3Cz14u6q

Hopefully, it will be okay soon

Scandal around dYdX over $25M

- DEX dYdX closed the bridge for ethDYDX migration — the native network since June 13.
- According to influencers, 45k+ holders are "stuck" with tokens worth over $25 million.
- Now ethDYDX cannot be exchanged, deposited to CEX, or converted.

Source

How Cross-Chain Bridges are Hacked?

Link: https://officercia.mirror.xyz/IvG5yxJrLviq0bT9CXMx8lQ-ZGOBomYbeizMEzp6n40

Two malicious crates have been identified in the crates.io

https://blog.rust-lang.org/2025/09/24/crates.io-malicious-crates-fasterlog-and-asyncprintln/

In Kazakhstan, the largest crypto exchange that worked for the drug trade was closed

The service was considered «respected» in the underground environment and cooperated with 20 largest «Darknet»-marketing, where total audience exceeded 5 million users. More than 200 drug trafficking sites from Kazakhstan, Russia, Ukraine and Moldova passed through it.

The total turnover of «RAKS exchange» exceeded 224 million USD.

https://sozmedia.kz/94819/

State of MCP Server Security 2025: 5,200 Servers, Credential Risks - https://astrix.security/learn/blog/state-of-mcp-server-security-2025/

We analyzed over 5,200 unique, open-source MCP server implementations to understand how they manage credentials and what this means for the security of the growing AI agent ecosystem.

- 88% of MCP servers need credentials to function
- 53% rely on static API keys and Personal Access Tokens (PAT)
- Only 8.5% use modern OAuth authentication
- 79% store API keys in basic environment variables

#MCP #ModelContextProtocol #AIAgents #AgentSecurity #CredentialSecurity #SecretsManagement #OAuth #APIKeys #PATs #SecretRotation #LeastPrivilege #AstrixSecurity

A very detailed Balancer hack post-mortem (unofficial): https://x.com/officer_secret/status/1985961846805843984?s=46

#security

📌 LLM Vulnerabilities: how AI apps break — and how to harden them

This piece maps the most common ways LLM-powered systems fail in the real world and turns them into a practical hardening plan. From prompt and indirect injection to over-privileged tools, leaky RAG pipelines, data poisoning, jailbreaks, and supply-chain traps — plus the guardrails that actually help in production.

Key takeaways:
1️⃣ Prompt & indirect injection: attackers hide instructions in web pages, files, or retrieved notes; the model obeys them and exfiltrates secrets or performs unwanted actions.
2️⃣ Jailbreaks & policy evasion: harmless-looking reformulations bypass safety layers; outputs become unsafe or operationally risky.
3️⃣ RAG data leaks: sloppy retrieval exposes internal docs, customer data, and system prompts; cross-tenant bleed is a real risk.
4️⃣ Over-privileged tools/agents: broad filesystem, network, or payment permissions turn one prompt into a breach.
5️⃣ Poisoning & supply chain: tainted datasets, third-party prompts, and unpinned models/extensions undermine trust.
6️⃣ Output trust & hallucinations: fabricated facts sneak into workflows, tickets, or code — and humans often rubber-stamp them.
7️⃣ Telemetry gaps: without red-team sims and runtime monitoring, you won’t see injection attempts until damage is done.

👉 Read here: https://netlas.io/blog/llm_vulnerabilities/

be careful)

🚨 ALERT: A fake Hyperliquid app has appeared on the Google Play Store, according to ZachXBT.

News | Markets | YouTube

12 LLM CTFs & Challenges - https://taleliyahu.medium.com/llm-ctfs-challenges-03dd55a9b7e4

Hands on CTFs and labs for LLM security. Train on prompt injection, jailbreaks, guardrail bypass, tool and agent abuse, data leaks, model inversion, and MCP issues.

Worth a read

I Checked the Worst OpSec Practices So You Don’t Have To

Link: https://officercia.medium.com/i-checked-the-worst-opsec-practices-so-you-dont-have-to-59501825c50f