Twitter is not a real place. Social media can reveal trouble or an emerging issue, but it does not reliably measure public opinion. Angry posts may reflect a vocal minority or an organized campaign. Examine their source and reach. Use sound polling, interviews, and message testing. Volume alone does not establish consensus.

Beware of viral content. Nobody is immune to manipulation. Authenticate suspicious images and recordings before relying on them. Obtain originals, trace the source, and consult forensic specialists. Preserve evidence and report impersonation. Do not dismiss genuine evidence as a deepfake or accept a convincing fabrication as fact.

The humans must remain in charge. A person must take responsibility for all crisis communications. Supervise AI use and protect sensitive information. Check facts, sources, privacy, and tone before sharing. Pause inappropriate scheduled promotions and automated replies. Confidence is one of many things a machine can produce without evidence.

Credibility can’t be bought. Automated accounts and AI-generated personas must not masquerade as genuine public support. Enlist credible experts and trusted community voices online. Disclose paid relationships. Never invent supporters, fabricate reviews, or present purchased endorsements as independent judgment. Disclosure does not make fabrication honest.

Don’t blame the machine. You are accountable when your technology causes harm. Shut down or restrict the affected function, preserve records, and involve qualified experts. Explain what happened and how to seek review or relief. Check the system before restoring it. Hiring a vendor does not outsource responsibility.