Facebook announced yesterday its limited policy to ban deepfake videos but stopped short by not attempting to eliminate altered or ‘fake’ content beyond that, e.g. altered photos, videos and news reports. See Facebook’s Deepfake Ban is a Solution to a Distant Problem
You may also have read about Twitter’s Jack Dorsey’s project to try and decentralize social networks/media so that users control their own content feeds with their own chosen algorithms. Dorsey readily admits the problem of fake news/media is too big for any one company to ‘police’ and solve. See our last blog on solving fake news problems Can Web 3.0 and Blockchain save us from Fake News and Content?
This week we witnessed a glaring example of an allegedly altered photo and the kind of partisanship and polarization it fosters, pointing to the immediacy of this problem. A GOP Congressman Tweeted a Fake Image of Obama with the Iranian President; they Never Met
The problem of ‘fakes’ goes well beyond news and images/videos. It extends to food, electronics, pharmaceuticals, luxury goods, and just about anything else we consume physically or logically. I did an interview with ISMG’s Tom Field on this subject before the Holidays – in case you are interested in listening to it: Gartner’s Avivah Litan on the Rise of Fake Everything
Digital Provenance Verification – a Layered Approach
To digitally verify provenance, the best practice is a ‘layered security’ approach – i.e. whitelisting with blockchain and blacklisting/anomaly detection with AI/ML models and other detection models. This layered approach works best in enterprise security. It also works best in verifying provenance.
Multiple technologies should and can be utilized:
- Whitelisting: Blockchain recording for provenance and asset tracking
- Authentication: Other technologies such as electronic impulse testing or spectral imaging to ensure the authenticity of the article (e.g. news, food, electronics) being recorded and tracked on the blockchain
- Blacklisting or Anomaly Detection: using AI and machine learning models to detect behavioral or data anomalies for assets/information/activities both on and off the blockchain that feed into the blockchain
- Bridging Physical and Virtual Truth: IoT sensors/networks/digital twins to connect physical ‘truth’ to virtual business event ‘truth’ recorded on the blockchain
We are getting desperate for truth in a trustless world. Technology can certainly play a role on delivering it. One of our Gartner 2020 Predicts is how Blockchain and AI will combine to fight fake news and images/videos. See Predicts 2020: Blockchain Technology Hope we are right!