AI safety, explained in plain language. Independent, calm and written for everyone.
Sample contentPreview edition. Every story here is sample content written to demonstrate the design — not real reporting.
Alignment is the work of making sure an AI system pursues the goal people actually intended. Here is the idea without the jargon, and why it matters for the tools you already use.test
Read story04 essentials — start here
ScamsSample
Scammers can imitate a familiar voice from a short recording. Two simple questions, agreed in advance, are enough to stop most attempts.
4 min read
DeepfakesSample
Odd blinking and blurry edges used to give fakes away. As the tools improve, the smarter question is not “does it look real?” but “who posted it first?”
5 min read
PrivacySample
AI assistants can keep your conversations, and some build a running profile to personalise answers. Here is what that means and how to stay in control.
5 min read
AI AgentsSample
An AI agent can browse, fill in forms and complete tasks on your behalf. That is useful, and it means mistakes become actions rather than words.
6 min read
9 stories
4 min
ScamsSample
Scammers can imitate a familiar voice from a short recording. Two simple questions, agreed in advance, are enough to stop most attempts.
5 min
DeepfakesSample
Odd blinking and blurry edges used to give fakes away. As the tools improve, the smarter question is not “does it look real?” but “who posted it first?”
5 min
PrivacySample
AI assistants can keep your conversations, and some build a running profile to personalise answers. Here is what that means and how to stay in control.
6 min
AlignmentAI AgentsSample
An AI agent can browse, fill in forms and complete tasks on your behalf. That is useful, and it means mistakes become actions rather than words.
5 min
PolicySample
Most AI rule-making comes down to the same handful of goals: know the risk, test before release, tell people the truth, and name who is responsible.
3 min
ScamsSample
Scam emails are now fluent, polite and personalised. The reliable warning signs are about behaviour, not typos.
3 min
PrivacySample
Photos carry faces, places and sometimes hidden location data. Four quick questions help you decide what is safe to share with an AI tool.
4 min
PolicyTestingSample
Before powerful AI systems are released they are increasingly put through structured tests. Here is what those tests can and cannot tell us.
2 min
DeepfakesSample
You do not need special software to resist voice and video impersonation. You need a word that only your household knows.
Nothing on the wire for that.
No stories match . Try a different word or clear the filters.
