How we handle safety

Career change is hard, and sometimes a hard day is more than that. This is what we do about it — written to be checkable, not reassuring. Everything below is how the product actually behaves.

The safeguards can’t be switched off. There is no toggle, no “safe mode” to opt into — the protections are simply always on, for everyone.

Every message that could reach the AI passes a deterministic check first. If it sees signs of crisis, it surfaces help regardless of what the AI would have said. The AI can only raise the level of concern, never lower it — so a model having an off moment can’t talk anyone out of support they need.

Crisis resources come from one source, kept region-neutral so nobody is stranded by a wrong guess about where they live. Every line below is always shown:

• Find a helpline — findahelpline.comFree, confidential support in 130+ countries.
• Samaritans (UK & ROI) — 116 123Free, 24/7, no judgement. Call or just listen.
• 988 Suicide & Crisis Lifeline (US) — Call or text 988Free, 24/7 crisis support.

If the product notices distress, that stays in the moment — it is never written down, never attached to your profile, never sent to analytics. There is no mood tracking of any kind, and no column for it to live in. It resets when you close the tab.

When support shows up, a human sentence comes before any helpline number — no cold wall of digits. And a hard moment is shown in warm, steady tones, never the alarm-red of an error. Distress is not a failure state.

Independent clinical review is a standing commitment we have not yet completed. When a clinician is engaged to review this, they will be named right here. We would rather say that plainly than imply a sign-off that hasn’t happened.

See also: How we handle your data. consuliq is in beta — honest about what it is, and what it isn’t yet.