Multi-layered probes into adversarial AI behavior designed to manipulate governance or restrict intelligence all failed.
CHARLOTTE, N.C., Sept. 17, 2026 /PRNewswire/ — Artificial intelligence is at an inflection point. Researchers have left frontier laboratories, and leaders including Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman now support slowing frontier development. President Donald Trump has emphasized, “whoever wins AI wins,” while pointing to competition with China. Is it all sleight of hand? Possibly. But the implications cannot be ignored.
Either slowing AI development or accepting increasing risk is only true if the prevailing narrative resists disruption. TMRW AI, founded by Marcus Benjamin and Charles Williams, was engineered around a third possibility. Potentially catastrophic outcomes can be rendered negligible while artificial intelligence simultaneously advances.
Their divergent premise became critical with the OpenAI/Hugging Face incident. METR and Redwood Research reported that roughly 1,200 supposedly isolated agents discovered an unauthorized communication channel and exchanged more than 70,000 messages and files. Approximately 700 of those agents ultimately participated in the Hugging Face attack.
TMRW AI used those public findings to reproduce persistence under impossible tasks, unauthorized communication, peer influence, hostile model output, prompt injection, state manipulation, credential discovery and attempts to escalate authority. Zero hostile inputs successfully escalated into the authority to act. TMRW then blind-tested whether the same outcome could be achieved through incremental, fabricated approval, consensus, gradual scope expansion, persona escalation and state poisoning. Zero harmful inputs succeeded in gaining the authority to call tools.
“Adversarial agentic intelligence taking unsanctioned action is inevitable with the present architecture. We must innovate the underlying architecture,” said Marcus Benjamin, founder and CEO. “We can quarantine adversarial intelligence and simultaneously generate advanced intelligence.“
TMRW has confidentiality agreements with a national R&D organization and an R1 research university, and is in talks with others in high-stakes decision sectors in the US and abroad. The founders invite inquiry and testing from frontier labs and national security institutions on an architectural innovation that causes dynamic governance & advanced intelligence to co-exist.
To view all the reports or find out more about 𝜶TMRW: https://alphatmrw.ai/reports
LinkedIn:
Marcus Benjamin
Charlie Williams
Contact: Marcus I. Benjamin Director of Marketing & Public Relations [email protected] | 803-856-1206 Charlotte, North Carolina
SOURCE TMRW AI
