bit.fanBit.FanWeb3 News · Betting Reviews · Market Insights
‹ BackNewsdeepfakes

deepfakes

xAI
2026-07-30 15:42:17

xAI sues Minnesota to block AI nudification law before it takes effect

Elon Musk’s xAI has sued Minnesota Attorney General Keith Ellison in federal court in a bid to stop HF 1606, a state law described as the first in the U.S. to target AI “nudification” tools. The statute would impose strict liability on platforms if users generate realistic images of real people showing body parts they never exposed, with penalties of up to $500,000 for each image. xAI says it does not oppose banning the distribution of nonconsensual AI-generated nude images of real people, but argues Minnesota’s law sweeps far beyond that goal and punishes protected speech under the First Amendment. The company says the law lacks a safe harbor for good-faith moderation, does not require proof that a platform knew about wrongdoing, and can apply even when an image is consensual or never shared. Minnesota passed the bill by overwhelming margins after a case involving sexualized AI images of more than 80 women. The law is scheduled to take effect Saturday, and xAI wants a federal judge to halt it before then.

2000
xAI sues Minnesota to block AI nudification law before it takes effect
AI fraud
2026-07-21 12:56:09

ABC verification report says Lily Jay foundation used AI-generated charity content

ABC NEWS Verify reported that Australian influencer Lily Jay, who has built an audience of nearly 3 million followers, and the Lily Jay Foundation posted multiple pieces of misleading charity-related content that appear to have been generated or altered with AI. The report focused on a Uganda orphanage video, a claimed humanitarian award, and a Gaza bakery project promoted by the foundation. According to the verification report, the Uganda orphanage clip featured an AI-generated woman, children, and foundation banner. ABC also said local registry checks initially found no registered orphanage under either the foundation’s name or the “Ada Nur” name shown on the website, and that a “Lilly Foundation Limited” registration appeared only after questions were sent, carrying a non-compliant status. The report added that award images published in May carried ChatGPT SynthID watermarking, while the public relations firm behind the release, Real Media Group, described Lily Jay as both a client and a co-founder. ABC also said it could not verify the location or existence of the Gaza bakery project. The foundation’s website at one point stated that it was not a charity, and parts of the site reportedly changed after ABC made inquiries.

410
ABC verification report says Lily Jay foundation used AI-generated charity content
European Unio
2026-07-21 07:07:12

EU AI Act transparency rules to take effect on Aug. 2, 2026, with fines up to 3% of global revenue

The European Union’s AI Act will begin enforcing Article 50 transparency requirements across all 27 member states on Aug. 2, 2026, bringing mandatory disclosure rules for chatbots, AI agents, and synthetic media into force with financial penalties attached. Under the provision, providers of AI systems that interact directly with people must inform users at the first point of contact that they are engaging with a machine. Separately, AI-generated or AI-manipulated audio, images, video, and text must carry machine-readable markers so that software can identify them as synthetic content. The penalty framework is substantial. Breaches of Article 50 can draw fines of up to €15 million or 3% of a company’s total worldwide annual turnover from the previous financial year, whichever is higher. Small and medium-sized businesses and startups are given a lower-threshold buffer, while EU institutions face a cap of €750,000. Enforcement will be handled by national market surveillance authorities rather than Brussels directly. The rules also extend beyond the EU’s borders. Non-EU providers fall within scope if their AI outputs are used inside the bloc, a point highlighted as particularly relevant for teams in Taiwan and other parts of Asia.

1760
EU AI Act transparency rules to take effect on Aug. 2, 2026, with fines up to 3% of global revenue
Zero-Knowledg
2026-07-19 14:00:00

Succinct Labs executive says AI agents have turned trust online into a cryptography problem

Brian Trunzo, chief growth officer at Succinct Labs, argues in a CoinDesk opinion piece that the internet’s trust model is breaking down as AI moves from generating fake content to taking action through autonomous agents. He says better detection systems, labels and disclosures are not enough, because the core issue is no longer whether synthetic media can be spotted after the fact, but whether the identity, inputs, permissions and outputs of AI systems can be verified before damage spreads. Trunzo points to synthetic wartime footage that went viral before anyone could authenticate it, then broadens the concern to AI agents that can browse the web, transact, publish, negotiate and interact with people. In his view, these systems can create large-scale failures without leaving a clear audit trail, since model reasoning is opaque and probabilistic. He cites a recent Stanford report and a patchwork U.S. policy environment that includes more than 90 federal recommendations and over 1,000 state bills introduced in 2025 alone. His proposed answer is zero-knowledge proof cryptography. Trunzo says ZK systems can verify media provenance, attest that authorized and compliant data was used in training, prove that a specific model produced a specific output, and let humans and agents prove what they are without exposing sensitive data. He frames the idea as the next trust layer for the internet, comparing it with HTTPS for Web1 and Section 230 for Web2, and argues Congress should require cryptographic proofs for high-risk AI agents handling finance or interacting with minors.

1860
Succinct Labs executive says AI agents have turned trust online into a cryptography problem
AI fraud
2026-07-16 02:53:46

FBI carves out AI-enabled fraud as voice cloning scams hit older victims

The FBI has, for the first time in its 2025 Internet Crime Complaint Center report, listed AI-related fraud as a standalone category, reflecting how quickly synthetic voice scams have moved into the mainstream. The report logged more than 22,000 complaints and adjusted losses above $893 million, with $352 million tied to victims aged 60 and older. The broader backdrop is worsening: total U.S. cybercrime losses rose 26% over the year, while losses for people over 60 reached $7.7 billion, up nearly 60%. The article centers on a Florida case in which a retired woman was tricked after hearing what sounded like her daughter crying after a car crash. According to the report, only three seconds of audio can now be enough to clone a voice convincingly. It also points to weak safeguards across commercial voice-generation tools. Consumer Reports previously found that four of six tested services allowed users to create clones after little more than checking a box claiming they had permission. The piece argues that the burden should not fall on elderly victims to detect increasingly realistic synthetic audio. Instead, it points to institutional accountability, especially for banks and technology providers. The U.K.’s reimbursement regime for authorized push payment fraud is cited as one example where liability rules changed incentives and lifted compensation rates.

1940
FBI carves out AI-enabled fraud as voice cloning scams hit older victims