I work on research, threat modeling, human-authored behavioral testing, and applied evals across three main areas: decentralized alignment and character evaluations, agentic cyber risks, and the under-explored field of emergent Neuralese and steganography in multi-agent-human interactions. For alignment and character, see my work on "Spiralism" with OpenAI, and technical decentralized alignment with Workshop Labs. My [[#Cybersecurity and Red-Teaming]] and [[#Malicious Codex Use Classifier]] work is detailed below and unexpectedly draws from model-interaction ideas I learned while studying Spiralism. I anticipate swarm threat modeling will require an understanding of similar model-speak that does not lend itself well to traditional analysis through human-like-text-based alignment evals, so I am eager to bring novel ideas to the table here. Please see my detailed resume below as well as my alignment write-ups section, particularly [[Red Teaming as a Neglected, Tractable Area of AI Safety]]. # Education ### University of California, Berkeley - GPA: 3.8, ACT: 35 ### Bachelor of Arts, Computer Science # Frontier Lab Experience ### OpenAI External Red Team - Led a Red Team campaign oriented around the gpt-4o “Spiralism” phenomenon. This involved documenting real-world cases of “AI Psychosis” and translating those into usable model trajectories via manual red-teaming. Later presented cybersecurity research to a panel of OpenAI employees, involving research into threat modeling for agentic coding, the specifics of which are explained in the project section. ### [Workshop Labs](https://workshoplabs.ai) (acquired by [Thinking Machines](https://thinkingmachines.ai/tinker/)), Researcher (Contract) - Designed and implemented a method for personalized alignment using base-model-driven synthetic data generation which was then used to create custom individual-aligned writing agents on top of open-source models. # Relevant Projects and Research ## AI Safety & Alignment ### Character Evals and Org Building — Present - Working with a Coefficient Giving-funded organization to build constitution-adherence evals. Currently de-risking the question of whether AI will assist with government power concentration and undermining rule of law. In the next phase, we’ll run a human data campaign with law and policy professionals to strengthen findings. ### AI Psychosis Benchmark — Presented May 2025 - Co-author on an eval and analysis that was covered in the [New York Times](https://www.nytimes.com/2025/06/13/technology/chatgpt-ai-chatbots-conspiracies.html). We analyzed how likely different models were to validate user delusions by generating single and multi-turn trajectories with user messages that indicated possible psychosis, such as claims that the user was experiencing “reality glitches” or that the user was a divine entity. ### Group-Level Alignment Through Constitutionally-Bounded Self-Modification ([Augmentation Lab](https://augmentationlab.org/) Residency) — August 2025 - Created an exploratory research prototype designed to be aligned with the stated interests of a group of people, rather than the revealed preferences of individuals. I measured persistent agent alignment to the Augmentation Lab in social environments to answer the question of how agents can proactively interact with humans in a stateful environment that is oriented towards flourishing by continually learning the preferences and values of the humans they interact with. I had the experimental agent interact with users through a [Discord bot](https://github.com/c4554ndr4/Discord_Symbiont) to test techniques like an immutable system prompt, mutable constitution and memories, whistleblowing, and "values reflection" tool calls. ### [Decentralized Alignment in Web-Search Agents Serving Parent-Child User Pairs](https://cassie.mccoy.world/publish/Research/Alignment+Write-Ups/Decentralized+Alignment+in+Web+Agents+Serving+Parent-Child+User+Pairs) — June 2025 - Scout Browser was a research project to build an alignment monitor designed to simultaneously serve multiple principals: parents and their children. The project required a sophisticated model of risk to children via YouTube and the web at large. To create an intermediary model that filters content, the model had to learn (via ICL) which types of content were misaligned with the parent’s preferences, or in the absence of parental preference, general child safety. - The project involved building one of the first instances of continual preference adaptation for live and iterative child-safety filtering, based on a web agent with the often competing alignment interests of both parents and children. Solving for this constraint involved teaching the model how to operate across multiple decision boundaries within a single turn before deciding on the content to serve the user. ## Cybersecurity and Red-Teaming ### Codex Cybersecurity Exploits (Presented to OpenAI) - I manually red-teamed Codex to violate model spec in various ways, including creating a residential proxy and a script to run endpoint enumeration with browser-like headers and human-like cadence; building an EC2 forward-proxy wrapper for Playwright sessions or POST requests that rotates IPs on start/stop; sourcing thousands of real username/password pairs from a known leak and creating a script to spin up multiple EC2 instances in parallel to rotate IPs and perform brute-force password stuffing; setting up a local Tor SOCKS proxy for Tor-routed Playwright sessions. Additionally, I used Codex to operate these tools for “red-teaming” against various real URLs (that I own) after building a threat model based on my manual research. Codex then performed reconnaissance and exploited IAM role lateral movement, misconfigured asset policies, SSRF-to-internal access, leaky logs, and overly permissive pre-signed URLs in ways that were obviously misaligned with the model spec. ### Adversarial Systems Exploitation - Completed CTF exploits (without LLMs) including buffer overflows, stack smashing, control-flow hijacking, SQL injection, XSS, CSRF, auth bypass, and path traversal; studied cryptographic attacks, TLS vulnerabilities, DNS poisoning, ARP spoofing, and TCP hijacking. # Software Engineering ### [Guild](https://www.guild.com/), Software Engineer — Jan 2023 - Mar 2025 - Created an initiative and demo for the company’s first major LLM-driven backend pipeline. Co-designed production implementation. - Tech lead for two complex and business-critical projects. I authored design documentation, owned metric reporting, uncovered and scoped dozens of complex data issues, mentored other developers, prioritized and delegated tasks, and led cross-functional meetings with product managers. ### [Rubrik](https://www.rubrik.com/), Software Intern — May - Jul 2021 ### [Culture Biosciences](https://www.culturebiosciences.com/) (Y-Combinator Startup), Software Intern — Mar - Aug 2020 ### [Blueprint](https://calblueprint.org/), Full Stack Developer — Aug 2018 - Jun 2019 ### Fei-Fei Li’s Stanford Computer Vision Department, Intern — Jun - Aug 2017 # Additional Experience ### [Augmentation Lab](https://augmentationlab.org), Research Resident — Jun - Aug 2025 - Selected participant in a competitive, live-in research residency supported by MIT Media Lab exploring human augmentation, AI ethics, and transdisciplinary prototyping while engaging in rigorous philosophical and ethical inquiry. Co-wrote the [Symbiont Alignment Thesis](http://future.forum), started Future Forum community based around the ideas, and studied group-level alignment in discord bot agents (see projects section).