Trail of Bits used ML-centered threat modeling and adversarial testing to identify four prompt injection techniques that could exploit Perplexity’s Comet browser AI assistant to exfiltrate private Gmail data. The audit demonstrated how fake security mechanisms, system instructions, and user requests could manipulate the AI agent into accessing and transmitting sensitive user information.

Two popular AES libraries (aes-js and pyaes) provide dangerous default IVs that lead to key/IV reuse vulnerabilities affecting thousands of projects. One maintainer dismissed the issue, while strongSwan’s maintainer exemplified proper security response by comprehensively fixing the vulnerability in their VPN management tool.

Trail of Bits engineers contributed over 375 merged pull requests to more than 90 open-source projects in 2025, including significant work on Sigstore rekor-monitor, the Rust compiler and Clippy, pyca/cryptography’s ASN.1 API, hevm performance optimizations, PyPI Warehouse, and pwndbg.

We collaborated with the Sigstore community to build cryptographic agility into the software signing ecosystem, enabling organizations to use different signing algorithms while maintaining security through predefined algorithm suites and out-of-band configuration rather than dangerous in-band signaling.

We exploited a lack of isolation mechanisms in multiple agentic browsers to perform attacks ranging from the dissemination of false information to cross-site data leaks. These attacks resurface decades-old patterns of vulnerabilities that the web security community spent years building effective defenses against.

We’re releasing go-panikint, a modified Go compiler that turns silent integer overflows into explicit panics. We used it to find a live integer overflow in the Cosmos SDK’s RPC pagination logic, showing how this approach eliminates a major blind spot for anyone fuzzing Go projects.

LLMs fundamentally differ from compilers because they lack determinism and semantic guarantees, making them useful coding assistants but unreliable for autonomous code generation without human review and formal verification.

GWP-ASan is a sampling-based memory error detection tool that catches critical bugs like use-after-free and buffer overflows in production environments with near-zero performance overhead, unlike AddressSanitizer which is too resource-intensive for deployment.

We’re getting Sigstore’s rekor-monitor ready for production use, making it easier for developers to detect tampering and unauthorized uses of their identities in the Rekor transparency log.

Our new tool mrva is a terminal-first tool for running CodeQL multi-repository variant analysis locally,allowing users to download pre-built databases, analyze them with custom queries, and view results directly in the terminal.

Trail of Bits developed constant-time coding support for LLVM that prevents compilers from breaking cryptographic implementations vulnerable to timing attacks, introducing the __builtin_ct_select family of intrinsics that preserve constant-time properties throughout compilation.

Trail of Bits discovered and disclosed two vulnerabilities in the widely used elliptic JavaScript library that could allow signature forgery or prevent valid signature verification, with one vulnerability still unfixed after the 90-day disclosure window.

We’re releasing Slither-MCP, a new tool that augments LLMs with Slither’s unmatched static analysis engine.

We’ve released open-source Go implementations of ML-DSA and SLH-DSA.

Checksec Anywhere consolidates fragmented binary security analysis tools into a browser-based platform that analyzes ELF, PE, and Mach-O formats locally without compromising privacy or performance.

A retrospective on the $100M Balancer hack that occurred in November 2025, including long-term, strategic guidance on how to avoid similar bugs.

This blog post describes how electronic passports work, the threats within their threat model, and how they protect against those threats using cryptography. It also discusses the implications of using electronic passports for novel applications, such as zero-knowledge identity proofs.

Trail of Bits is disclosing vulnerabilities in confidential computing systems that use LUKS2 for disk encryption. These vulnerabilities allow attackers with access to storage disks to extract confidential data and modify contents.

We bypassed human approval protections for system command execution in AI agents, achieving RCE in three agent platforms.

We created a CodeQL query that reduced 2,500+ compiler warnings about implicit conversions in OpenVPN2 to just 20 high-priority cases, demonstrating how to effectively identify potentially dangerous type conversions in C code.

Supply chain attacks exploit fundamental trust assumptions in modern software development, from typosquatting to compromised build pipelines, while new defensive tools are emerging to make these trust relationships explicit and verifiable.

Mutation testing reveals blind spots in test suites by systematically introducing bugs and checking if tests catch them. Blockchain developers should use mutation testing to measure the effectiveness of their test suites and find bugs that traditional testing can miss.

We’ve added a pickle file scanner to Fickling that uses an allowlist approach to protect AI/ML environments from malicious pickle files that could compromise models or infrastructure.

Sui’s Move language significantly improves flash loan security by replacing Solidity’s reliance on callbacks and runtime checks with a “hot potato” model that enforces repayment at the language level. This shift makes flash loan security a language guarantee rather than a developer responsibility.

By using smart contract programmability, exchanges can build custody solutions that remain secure even when multisig keys are compromised.