Security Spotlight: Navigating AI at Work: Partner or Threat?
AI enables an enormous gain in productivity, but it cannot replace your common sense. By questioning your habits and remaining vigilant, you can significantly reduce risks in your daily work and shrink the attack surface for cybercriminals.
Risks:
|
Data Privacy and Data Leakage |
AI servers use data for training which can’t be removed once stored. |
|
Autonomous AI Assistants
|
If autonomous AI assistants fail to shut down reliably, they can record and forward confidential follow-up conversations. |
|
Hallucinations
|
Language models input missing information which can lead to incorrect medical treatment. |
|
Bias
|
If AI training data contains biases, it will adopt and reinforce those patterns, often imperceptibly. |
|
Insecure Code |
AI code does not always meet standard security requirements. |
Protection:
|
Practice Zero Trust |
Always treat unexpected request skeptically, especially urgent messages. |
|
Verify Via Known Channels |
Verify using another method of communication. |
|
Minimize Data Sharing |
Never enter sensitive information into public AI systems. |
|
Check Settings |
Opt out of data usage for AI training; do not use unverified prompt templates. |
|
Critically Question Results |
Check content, reports, and code before you use them. |
|
Flag Anomalies |
Immediately report suspicious incidents to your Security Team. |