Pictures are from the Internet

Pictures are from the Internet

Pictures are from the Internet

Can you believe the "AI face swap" in the hot searches, is it true?

"Friends, guess who this video is?" - When you receive a private message from an "old classmate" on a social platform, click on the video and find that the other person's face is actually a celebrity or even an amateur, are you scared at this time?

In March 2024, the financial director of a company in Guangdong was cheated out of 8.7 million yuan for an AI face-changing video of the "boss"; in the same month, the number of readings on Douyin's AI face-changing fraud topic exceeded 5 billion times. Behind these numbers, an ethical crisis caused by AI technology is sweeping the world. As self-media people, we have to ask: When technological innovation becomes an accomplice of crime, who should pay for this "perfect crime"?

Pictures are from the Internet

1. AI face-changing scam: a "Trumen Show" that subverts trust

Case direct hit:

In July last year, 23-year-old Hangzhou college student Xiaolin was lent in a QQ group with a "senior sister" who borrowed money. The other party used an AI face-changing video to prove his identity, and eventually cheated his parents of 200,000 yuan in savings. What’s even more terrifying is that the police investigation shows that the cost of producing such fraudulent videos is less than 30 yuan!

Pictures are from the Internet

Technical Anatomy:

The current AI face swap technology has achieved "generating faces in 5 seconds completing face swaps in 3 minutes", and the number of downloads of related projects in the open source code library exceeds 1.2 million times. A certain review shows that 95% of people cannot realize the face-changing video in 2.0 version in 10 seconds.

Pictures are from the Internet

Pictures are from the Internet

In-depth reflection:

Trust collapses: When faces can be forged, how will the identity authentication system in human society be reconstructed?

Technology black market: A dark web platform priced at 2,999 yuan has a monthly sales of 12,000 copies, and the regulatory vacuum is astonishing!

Pictures are from the Internet

Pictures are taken from the Internet

Legal dilemma: The current Criminal Law "the crime of forging documents" is difficult to cover AI face-changing, and victims often face the dual desperate situation of "difficulty in providing evidence and difficulty in recovering stolen goods".

Pictures are from the Internet

Pictures are from the Internet

2. AI "virtual girlfriend" attracts traffic: the gray industrial chain of short video platforms

Phenomenon scan:

Open Douyin, the number of views of tags such as "AI Miss Chat with You" and "Virtual Idol with 24-hour live broadcast" exceeded 50 billion times. A MCN institution revealed that they generated 1,000 "Internet celebrity" accounts through AI, and their monthly advertising revenue exceeded 8 million yuan.

Pictures are from the Internet

Pictures are from the Internet

Chain of interests:

1. Data feeding: The preference data generated by users when watching videos is used to train an AI image that "knows you".

2. Traffic harvesting: Use algorithms to push AI accounts to target users, and the conversion rate is 3-5 times that of real-person IP.

3. Emotional kidnapping: An AI anchor induces fans to recharge 688 yuan to purchase "exclusive skin", which was later confirmed to be false marketing.

Pictures are from the Internet

Pictures are from the Internet

Ethical questioning:

Fake Personas: When AI can imitate anyone's voice and expression, how do we distinguish between reality and illusion?

Labor Exploitation: Behind an AI singer are the "voice slaves" of 300 live voice actors, and they record the audio every minute for 0.3 yuan in return.

Cognitive pollution: The "perfect image" generated by adolescents' long-term exposure to AI may lead to self-identity crisis and aesthetic alienation.

Pictures are from the Internet

Pictures are from the Internet

3. Behind Lei Jun’s proposal: The awakening of responsibility of technology giants

Focus of the Two Sessions:

During the 2024 National People's Congress and the Chinese People's Political Consultative Conference, Lei Jun submitted the "Suggestions on Strengthening the Governance of AI Face Change Technology", which clearly proposed seven measures, including "establishing an AI face recognition hierarchical management system" and "requiring video platforms to deploy an AI face change detection system".

Pictures are from the Internet

Industry response:

WeChat has launched the "Video Number AI Face Change Detection" function, with an accuracy rate of 92%.

Bilibili marked the "virtual creation" watermark on the content generated by AI, but there is still a 30% missed detection rate.

An AI company launched an "ethical firewall" that claims to intercept 90% of malicious AI application requests.

Pictures are taken from the Internet

Voice of controversy:

Technical Optimists: "Overregulation will hinder innovation and should be left to the market to solve it"

Security faction: "It will be too late to legislate after large-scale fraud breaks out, and it is urgent to refer to the European Union's Artificial Intelligence Act."

Public anxiety: How much defensive capabilities do ordinary people have in the torrent of technology?

4. AI creation copyright war: Who is the real "creator"?

Typical cases:

In 2023, a painter sued the AI ​​painting website, accusing him of using his works to train models without authorization and claiming 120 million yuan. On the other hand, the AI-generated online novel "Urban Cultivation" was shortlisted for a shortlist of a literary award, causing heated debate in the industry.

Pictures are from the Internet

Legal blind spots:

International comparison: The US Copyright Office explicitly refuses to register AI-generated works, while the UK recognizes the "limited copyright" of AI-generated works

Current situation in China: The current Copyright Law protects works with "original expression", but it is still unclear whether AI-generated content meets the requirements

Industry dilemma: Data from a music platform shows that the proportion of background music generated by AI has increased from 5% to 23%, and the income of original musicians has dropped by 37%.

Pictures are taken from the Internet

Pictures are from the Internet

Philosophical Question:

If AI can write a masterpiece like "A Hundred Years of Solitude", what is the value of a human author?

Pictures are taken from the Internet

When the generated pictures are more exquisite than the photographer, where is the boundary of photography art?

Will technological progress inevitably lead to "art democratization" or create a new class gap?

Pictures are taken from the Internet

Standing at the intersection of AI ethics, we are all "competitors".

From deep pseudo videos to virtual idols, from algorithmic hegemony to copyright wars, AI is reshaping the underlying logic of human civilization. As self-media people, we are both beneficiaries of technology and gatekeepers of the truth.

Pictures are from the Internet

Pictures are taken from the Internet

Here we call:

1. Platform: Establish a "red and black list" of AI content, making technology good a business instinct

2. Legal circle: Accelerate the formulation of the "Artificial Intelligence Ethics Review Law" and establish a "lifelong accountability system for technology abuse"

3. Public: Popularize AI literacy education and cultivate "critical thinking" as a sword to resist illusion.

Pictures are taken from the Internet

Finally, I will leave you a multiple-choice question:

When AI technology develops to perfectly replicate your personality, voice, and even memory, you want it to be:

A) "digital clone" across time and space

B) "Sky Eye System" to prevent fraud

C) The "ultimate opponent" that makes you unemployed

D) "Philosophical Tool" to Redefine Human Nature