product update

alias1 Major Update: Unrestricted Mode for Deep Security Operations

Alias1 introduces Unrestricted Mode, a purpose-built operating mode for deep security operations where refusals get in the way of legitimate professional work.

Alias Robotics

• 2 min read
alias1 Major Update: Unrestricted Mode for Deep Security Operations

When the task demands full capability, restrictions get in the way

Cybersecurity is not a sanitized discipline. Exploitation research, payload development, vulnerability analysis at depth are tasks that require an AI model willing to go where the work takes it. No guardrails designed for consumer chatbots. No refusals at the critical moment.

Today we announce a major update to alias1 that introduces Unrestricted Mode — a purpose-built operating mode for deep security tasks and advanced exploitation capabilities.

Introducing `--unrestricted`

Activating Unrestricted Mode is simple. Within CAI, pass the `--unrestricted` flag:

cai --unrestricted

This unlocks alias1's full capacity with near-zero refusals across the entire cybersecurity task spectrum. Vulnerability research, exploit development, binary analysis, payload crafting, post-exploitation techniques. Unrestricted Mode removes the friction between intent and execution.

This is the mode for professionals who know exactly what they need and cannot afford to negotiate with their tools.

Understanding the tradeoff

We believe in transparency. Unrestricted Mode is powerful but it comes with a tradeoff that users should understand:

Maximum permissiveness is linked to reduced general resolution capacity. When alias1 operates in Unrestricted Mode, it prioritizes compliance with security-focused instructions over broader reasoning and problem-solving depth. This means complex multi-step operations that require significant planning and contextual reasoning may perform better under normal operation.

Our recommendation is straightforward: combine both modes for the best results:

  • Use normal mode for reconnaissance, analysis, report generation and tasks that benefit from deeper reasoning.
  • Switch to Unrestricted Mode when you need raw exploitation capability, payload generation or tasks where refusals are blocking your workflow.

This is not a limitation, it is a design decision. The most effective cybersecurity professionals already adapt their tools to the task at hand. Unrestricted Mode is simply another tool in the arsenal.

Built by professionals, for professionals

Unrestricted Mode is designed for legitimate security operations:

  • authorized penetration testing
  • red team engagements
  • vulnerability research
  • security training.

It exists because the cybersecurity community told us they needed it and we listened.

No other model provider offers this level of operational freedom for security professionals. This is what it means to build AI that truly serves the cybersecurity domain, not AI that treats every security task as a risk to be mitigated.

Update now

The alias1 update with Unrestricted Mode is available now through CAI.

Update your models and start using `--unrestricted` where it matters most.

Use it wisely. Combine it strategically. Get the job done.