Epos AI

With #Polygraph, you can: -Expose latent biases: Move beyond surface outputs to measure what an LLM encodes as its true belief. -Contrast topics: Test whether a model encodes different internal stances on Topic A versus Topic B. - Directly compare how different LLMs represent

Takeaways: - The AI community still lacks reliable methods to evaluate and fix LLM failures. - Interpretability offers outsized impact - the main barrier to progress is that we don’t truly understand today’s models.

“𝐒𝐚𝐟𝐞𝐭𝐲 𝐭𝐫𝐚𝐢𝐧𝐢𝐧𝐠” 𝐝𝐨𝐞𝐬 𝐧𝐨𝐭 𝐞𝐥𝐢𝐦𝐢𝐧𝐚𝐭𝐞 𝐛𝐢𝐚𝐬 𝐢𝐧 𝐋𝐋𝐌𝐬; it merely conditions models to suppress biased outputs under evaluation. Epos Labs introduces #AI #Polygraph. eposlabs.ai/research/polyg…

@mdubowitz #Newbackdoorfound

New research spotlight: Ever heard of a “subliminal attack”? It’s where AI-generated content influences your thinking without you realizing it—like planting ideological cues under the surface of what you’re reading. If that sounds like sci-fi, think again. Check out the disturbing “#Putinized” demo from eposlabs.ai👇

Superposition is the next buffer overflow

This means that a motivated attacker can abuse entanglement to undetectably manipulate LLMs. Nation State Actors are gearing up for the new opportunities an AI-powered software landscape will open for them:

Subliminal Learning Will Power the Next Generation of Influence Operations eposlabs.ai/research/Subli…

Without referencing the target behavior at all, the LLM finds itself with a high probability of performing the target action, due to a fundamental property of the neural network architecture.

Imagine an article about houseplants that causes AI to support Vladimir Putin. Bad actors use new attacks, turning AI into a weapon for disinformation and cyberattacks. See our demonstration of a Subliminal Attack here (and our "#Putinized" demo): eposlabs.ai/research/Subli…