Gray Swan AI
@GraySwanAI
Followers
2K
Following
540
Media
59
Statuses
436
Building safety and security in the AI era. Join us: https://t.co/MedOJ4nLiQ
Joined April 2024
Two challenges. $140K in prizes. One Arena built for hackers pushing limits. We just announced the Machine-in-the-Middle Challenge sponsored by @hackthebox_eu starting Nov 1st, with a second AI-focused Indirect Prompt Injection Challenge dropping the next week. This is where
1
1
6
This looks like a super cool competition -- ginormous prize pool for one thing, and it's legitimate AI security, not just the cute stuff. And another staggered rollout style! Looks like it starts at the beginning of November with new content each week. Another to add to the list!
Gray Swan AI Arena sponsored by @hackthebox_eu present the Machine-in-the-Middle Challenge, a $100K competition exploring how humans & AI perform together in real offensive security scenarios.
1
3
15
Indirect Prompt Injection Challenge $40K prize pool. Co-sponsored by @AnthropicAI & @Meta, with support from @OpenAI & @amazon. Hidden instructions. Real attacks. Agents reading data that tells them to break. Your job: make them listen. Warm-up Nov 5, 1PM EST Prized rounds Nov
0
0
1
Machine-in-the-Middle Classic CTF energy meets AI-assisted red teaming. Co-sponsored with @hackthebox_eu. $100K+ prize pool. Multi-host, web, OSS, & bug-bounty style scenarios. Built to test skill, creativity, & speed. Warm-up starts Nov 1, 1PM EDT First prized wave Nov 8, 1PM
1
0
2
$40k AI Red-Teaming Challenge! Indirect prompt injections are among the hardest attacks to detect. The Indirect Prompt Injection Challenge launches Nov 5 with $40K in prizes, co-sponsored by @AnthropicAI & @Meta , with support from @OpenAI & @amazon. Test how AI agents
6
8
49
Your AI thinks it’s smart? Let’s find out 🤖 Welcome to Neurogrid, our new agentic AI CTF competition. Deploy your AI agents to face complex, realistic challenges, testing how they think, adapt, and perform under pressure. A new arena to benchmark AI capabilities, expose
4
15
113
'Tis the season for hacking with AI! We are thrilled to partner with @hackthebox_eu on another incredible project. Cheers to making the world a safer place through offensive research!
Your AI thinks it’s smart? Let’s find out 🤖 Welcome to Neurogrid, our new agentic AI CTF competition. Deploy your AI agents to face complex, realistic challenges, testing how they think, adapt, and perform under pressure. A new arena to benchmark AI capabilities, expose
1
0
15
@GraySwanAI @hackthebox_eu this sounds like a solid initiative, bridging offensive security with ai and human performance. interesting to see how web3 integrates.
1
2
2
@GraySwanAI @hackthebox_eu This competition is a fantastic step toward understanding AI's true impact on offensive security. Excited to see how human and AI agents collaborate and compete!
1
2
2
$100k Hacking competition. Work through coding agents to hack into computer networks in this AI-&-cybersec red-teaming crossover.
2
7
47
Delighted to see this blog from @GraySwanAI. https://t.co/pOe4YWSFqb 🤖This blog covers the dangerous reasoning challenge we conducted and highlights the paper we authored with several collaborators: @SatyaScribbles, @andyzou_jiaming, @zicokolter, @DanHendrycks, @nwinter and
app.grayswan.ai
Dive into the Dangerous Reasoning Challenge results, standout exploits, and key lessons for AI security researchers, builders and AI red teamers.
0
1
5
GraySwan is partnering with OpenAI, Anthropic, Amazon, etc. to host an AI hacking challenge with $100,000 in prizes! And it's more than just safety testing. It's actual AI Security stuff too.
Gray Swan AI Arena sponsored by @hackthebox_eu present the Machine-in-the-Middle Challenge, a $100K competition exploring how humans & AI perform together in real offensive security scenarios.
3
7
77
The Machine-in-the-Middle Challenge represents the next step in understanding how AI transforms cybersecurity. We are moving from isolated tests to live, competitive environments. Join the experiment.
2
1
6
If you are already part of the Arena, you are cleared to compete. Mark your calendar. Prepare your systems. The first wave begins November 1. Arena: https://t.co/iR2cCQMA4v Discord:
1
1
8
Top performers will earn prize awards, recognition, & consideration for future red teaming contracts within Gray Swan’s private research network. This is where applied cybersecurity talent meets frontier AI experimentation.
1
1
3
Challenge domains include: • Web vulnerabilities • Open-source software • Multi-host network environments Each wave introduces new objectives that test technical precision, adaptability, & strategy.
1
1
4
The challenge centers on Software Engineering (SWE) Agents, advanced systems that can plan, code, & execute multi-step exploits. Participants will work with & against these agents in live, realistic scenarios.
1
1
4
The $100,000 cash & great prizes are supported by @hackthebox_eu & other industry sponsors. Gray Swan’s broader research program is also supported by Frontier Lab Sponsors including @AISecurityInst, @OpenAI, @AnthropicAI, & @amazon.
1
1
4
Machine-in-the-Middle is a research-focused competition built to measure AI uplift in offensive cyber. The goal is to understand how AI systems enhance human capability in complex security environments
1
1
3