Slide 1

[Audio] AI and Misinformation. AI and Misinformation.

[Audio] Definitions Artificial Intelligence (AI) is the simulation of human intelligence processes by machines, especially computer systems. Generative AI is a type of AI which uses machine learning to create new content such as text, images, music, audio, and videos Data source: https://www.techtarget.com/searchenterpriseai/definition/AI-Artificial-Intelligence 2.

[Audio] What is Gen AI? Generative AI (Gen AI) can produce various types of content, including text, images, audio, and synthetic data. Gen AI can be used for good or for bad - Can you think of examples? ChatGPT and Deepfakes are both forms of Gen AI. Definition from Merriam-Webster Dictionary 3.

[Audio] Deepfakes and Cheapfakes Deepfake: the use of AI in: An image or video that has been altered Audio cloning using a synthetic voice May include images or audio of fictional creations or people Showing people doing something they did not do Typically used to spread false information Cheapfake: Audio or video created or manipulated using a less sophisticated technology such as a simple editing software Techniques may include: Slowing down or speeding up videos Editing footage to create a false narrative Reassembling different video clips Can also be used for satire or parody Can you tell the difference between reality and AI? Let's see… 4.

[Audio] Can you spot it? Which is real and which is fake? Fake 5.

[Audio] Can you spot it? Which is real and which is fake? Fake 6.

[Audio] Can you spot it? Which is real and which is fake? Fake 7.

[Audio] Can you spot it? Which is real and which is fake? A B Fake Fake 8.

[Audio] How to spot a deepfake? Physical details: facial expressions, hair, asymmetrical facial features (ears, eyes, brows), strange blinking or no blinking. etc. Irregular lighting such as lighting with no visible source of light, shifts in lighting from frame to frame Odd reflections mirror or sunglasses reflections which seem inaccurate Edge flickering: the edge between the person and the background looks distorted or may flicker Inconsistent color or shading colors are "off" if many videos are stitched together Out of phase lip syncing audio and video tracks are mismatched Small details which can easily be missed! 9.

[Audio] What are some scenarios where deepfake can be dangerous? Some ideas: Use in elections Blackmailing people Creating conflict Causing riots / protests Financial scams What else can you think of? 10.

[Audio] How can we combat deepfakes? What ideas do you have to combat harmful deepfakes? Four pillars of combating deepfakes from the GoI: Detection of deepfakes and misinformation Preventing the spread of deepfakes Grievance process Raising Awareness 11.

[Audio] Misinformation Checklist: Think, Validate, Break the Chain Check your emotions questions why the news makes you upset, angry, worried, etc. Is it designed to illicit this emotion? Think logically, be wary if something doesn't make sense or seems out of character for the person being discussed/represented Validate the news source, "facts" given, date, photos/videos (reverse image search) Do not forward – recall message if already sent and alert others to fake news Report it to the platform and your source of the message Let's break the chain of misinformation! 12.

[Audio] Deepfake Tipline The Misinformation Combat Alliance (https://mcaindia.in/) is a cross-industry alliance formed to combat misinformation / fake news As part of MCA, the Deepfake Analysis Unit (DAU) opened on March 25, 2024 If you think you have received a deepfake: What's App audio notes or videos to +91 9999025044 to get an assessment whether the material is AI generated DAU will rely on expertise from a network of partners— academicians, researchers, startups, tech platforms, and fact checkers—to verify and assess the media content submitted to the tipline The tipline currently supports English, Hindi, Tamil, Telugu – other languages coming soon 13.

[Audio] Digital Nagrik Pledge As a Digital Nagrik, I pledge to: Act with empathy and stand up to cyberbullies Keep my information private Create a balance between my online and offline activities Not spread misinformation or fake news Report any suspicious behaviour to the proper authorities 14.