Google's AI Breakthrough Tackles Hidden Deepfakes
The increasing sophistication of AI-generated content has raised concerns globally, as it becomes more difficult for both the public and experts to distinguish between real and fabricated videos. Deepfakes, which use AI to superimpose or manipulate faces, have already caused significant harm by spreading misinformation and misleading audiences. These video forgeries have been used in everything from political manipulation to creating harmful viral content.
Traditional deepfake detection systems primarily focus on facial recognition, scanning for anomalies or inconsistencies within the faces of individuals in videos. However, these methods fall short when the face is obscured or absent altogether, which is common in modern deepfake techniques. To address this gap, UNITE has been designed to look beyond faces, analysing the entire video frame to detect subtle visual cues that indicate manipulation.
ADVERTISEMENT
The system employs an innovative approach that involves scrutinising the background, motion patterns, and even the physics of the scene, which can reveal subtle discrepancies introduced by AI editing. For example, changes in the motion of objects, lighting inconsistencies, and unnatural background shifts are all telltale signs of deepfakes, and UNITE is trained to spot these signs with an exceptional degree of accuracy.
According to the researchers, the system works by comparing the video under scrutiny with a vast database of real-world footage, assessing what is considered natural or typical for specific environments, actions, and movements. These advanced methods are powered by deep learning algorithms that continuously improve their ability to identify fake content as they process more data.
The collaboration between UC Riverside and Google aims to make deepfake detection more accessible and efficient for various industries, including newsrooms and social media platforms. As misinformation continues to spread through manipulated content, news organisations are increasingly turning to automated systems to help verify the authenticity of video materials before publication. With UNITE, they gain a powerful tool that goes beyond face-based analysis to ensure the credibility of the content.
This AI solution is timely, given the surge in social media videos and user-generated content. Many platforms have struggled with moderating fake content, and deepfakes are becoming an especially potent tool for those seeking to spread disinformation. This is particularly concerning in political campaigns, where fake videos can be used to sway voter opinions or tarnish reputations.
UNITE's success could pave the way for more advanced technologies that can identify digital fakes across various forms of media, including text and audio, further protecting against the rising tide of fake content. The system also raises important questions about privacy and ethics, especially as it opens the door for new ways to track and monitor digital media for authenticity.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Al Etihad
2 hours ago
- Al Etihad
Google commits $1 billion for AI training at US universities
6 Aug 2025 23:13 SAN FRANCISCO (REUTERS) Alphabet's Google on Wednesday announced a three-year, $1 billion commitment to provide artificial intelligence training and tools to US higher education institutions and than 100 universities have signed on to the initiative so far, including some of the nation's largest public university systems such as Texas A&M and the University of North schools may receive cash funding and resources, such as cloud computing credits towards AI training for students as well as research on AI-related topics. The billion-dollar figure also includes the value of paid AI tools, such as an advanced version of the Gemini chatbot, which Google will give to college students for hopes to expand the programme to every accredited nonprofit college in the US, and is discussing similar plans in other countries, Senior Vice President James Manyika said in an declined to specify how much Google is earmarking in direct funds to external institutions relative to footing its own cloud and subscription announcement comes as rivals like OpenAI, Anthropic and Amazon have made similar pushes around AI in education as the technology pervades in July pledged $4 billion to bolster AI in education evangelising their products to students, tech firms further stand to win business deals once those users enter the workforce.A growing body of research has mapped concerns around AI's role in education, from enabling cheating to eroding critical thinking, prompting some schools to consider bans. Manyika said Google had not faced resistance from administrators since it began to plot its education initiative earlier this year, but "many more questions" about AI-related concerns remain. "We're hoping to learn together with these institutions about how best to use these tools," he said, adding that the insights could help shape future product decisions.
Campaign ME
7 hours ago
- Campaign ME
MENA industry reacts after Amazon abruptly exits Google Shopping Ads
Between July 21 and July 23, 2025, Amazon withdrew from virtually all Google Shopping auctions in major global markets, including the United States, the United Kingdom, and Germany. The dramatic move was first brought into sharp public focus by Mohammed Sajjad, Chief Marketing and Digital Officer at Azadea Group, who wrote, 'Amazon didn't reduce bids. Didn't trial a new format. They vanished from approximately 60 per cent of Google Shopping auctions. Overnight. Global. Coordinated.' This development follows a 50 per cent reduction in Amazon's Google Shopping ad spend in the United States in May 2025, suggesting that the July move was the culmination of a carefully orchestrated strategy rather than an impulsive decision. Market backdrop and competitive signals Amazon's withdrawal comes just as holiday shopping ramps up, with 71 per cent of consumers expected to begin before Thanksgiving, up 18 per cent year-over-year. The timing could significantly reshape traffic flows during the industry's most critical quarter, compelling competitors to adapt their campaigns earlier and more aggressively. This move also follows key shifts in Google's advertising ecosystem. In late 2024, Google removed Performance Max's priority over Standard Shopping campaigns, diminishing the dominance of automated ad placements. Simultaneously, Google Lens began serving shopping ads within its 20 billion monthly visual searches, and Perplexity AI entered the market with a one-click, AI-powered shopping assistant. In this context, Amazon's move is widely viewed as a deliberate shift away from paying for visibility on Google, and toward controlling the entire product discovery journey themselves. As Sajjad puts it, 'Amazon didn't tweak a line item — they redrew the map.' Industry leaders weigh in The advertising and e-commerce world responded with a wave of analysis and speculation. Executives and digital leaders interpreted the withdrawal as a calculated signal of Amazon's evolving position in the digital advertising ecosystem. Faheem Ahmad, Head of Growth at Chalhoub Group, framed the move as a data-driven optimisation play with broader implications for ad strategy, saying, 'They presumably spent $50m per year on shopping ads, given a 600 million inventory size. That's $230 per product per day, which is huge. Amazon may only continue advertising its private-label inventory, which represents just 2 per cent of its SKUs.' Asim Shaikh, eCommerce and Digital Leader at iHerb, framed the withdrawal as a structural shift rather than a tactical change. 'I believe this isn't about bid strategy, it's a pure platform power shift. Amazon recently blocked Google's 'Project Mariner' crawler. If Google's AI models train on your content, and you're paying them for traffic? That's a bad deal not only for Amazon but for most established retailers,' Shaikh said. Shaikh added that Amazon likely wants to be 'the first step in search, not the last click'. Strategic tension with Google Faisal Dean, CEO of Assembly's MENA region, contextualised Amazon's withdrawal as part of a broader trend in their relationship with Google. 'This is not the first time Amazon has pulled back spend. However, this seems to be the most drastic to date,' he noted. Dean also pointed to the growing contradiction in Amazon's continued funding of a competitor that is building directly competitive capabilities. 'The strategic contradiction of funding your competitor has become more apparent as Google's AI-powered contextual answers and zero-click queries could put Amazon at risk of over-reliance,' he said. Dean added that while the decision likely includes considerations around MMM (marketing mix modeling), LTV-driven incrementality testing, and seasonality-driven investments, there may also be a forward-looking element at play. He said, 'This could potentially be a future-facing strategy to stop part-funding, both with advertising dollars and data, Google's feature rollouts announced earlier this year, which aim to improve the personalised shopping experience and close the loop within Google itself i.e., stepping onto Amazon's turf, as well as free up millions in advertising spend to re-invest into their own growing retail media offering'. A data-driven power move According to current third-party estimates, Amazon Prime has approximately 220 to 260 million paying members globally, with around 180 million in the United States alone. This accounts for nearly 75–80 per cent of all US households. For many Prime subscribers, the purchase journey starts and ends on Amazon, making Google and other search platforms increasingly irrelevant. As Rasha Hamzeh, Managing Director at the Inhouse Agency, put it, 'Owning discovery, ditching dependency, not a test but a territory grab. Or maybe Amazon's betting that customers no longer need to be acquired. Not fighting for clicks but optimising for loyalty.' Darine Sabbagh, GM of E-commerce at Chalhoub Group, added, 'At their scale, there's enough critical mass in the zMOT journey to just 'Amazon it,' rather than 'Google it' first. Staying on Google may have simply been a way to crowd others out.' Chris Bishop, Ecommerce Director at SQUATWOLF, framed the withdrawal as a defining industry moment, saying, 'The king is dead, long live the king. Amazon stepping back from Google Shopping isn't just a gap in the auction – it may be a once-in-a-cycle shift in category dynamics.' Bishop adds, 'While Amazon tests how far it can pull back from funding Google's coffers without losing its grip on the customer, especially in territories where it has long commanded a disproportionate share of wallet, this could be an Ehrenberg-Bass moment: double down on penetration, win light buyers, and build mental availability while the giant is, perhaps arrogantly, asleep at the wheel.' What's next? With CPCs down 20–40 per cent, competitors are already capturing market share in Amazon's absence. Whether Amazon's self-sufficiency gamble pays off, or simply creates space for rivals to reclaim paid search real estate, remains to be seen. Bilal Adham, Group Vice President of Digital at DP World, sees this as a familiar pattern. Adham said, 'We've seen similar tactics from Shein, Temu, and Alibaba — flooding Google listing ads early on, then pulling back once they'd built a pool of users.' But in Amazon's case, the implications may run deeper. He added, 'If you take a step back, are Amazon telling us they've reached the holy grail of ecommerce? Scale acquisition, then shift to trust, experience, and habit. With app installs, repeat sessions, logged-in users, and CRM, it reduces the need for paid discovery altogether, playing on loyalty and advocacy.' Adham ended with a sharp reflection, saying, 'Back in 2014, Eric Schmidt called Amazon 'Google's biggest search competitor.' For Amazon, that moment may have quietly arrived.' For now, one thing is clear: Amazon has signaled it no longer intends to play by Google's rules. Whether others follow may define the next era of digital commerce. Authored by Nasser Oudjidane, Co-founder and CEO at
Tahawul Tech
8 hours ago
- Tahawul Tech
OpenAI, Google and Anthropic awarded US government contracts
OpenAI, Google and Anthropic have been given a significant boost following the decision by the US government's centralised procurement division to add the companies to its list of approved AI companies. That now paves the way for the tech companies to play central roles in the US government's anticipated widespread rollout of critical national use-cases of AI. Claude, Gemini and ChatGPT are the large language models that have been officially signed, sealed and delivered by the US General Services Administration (GSA) and will be subsequently be added to its Multiple Award Schedule (MAS) contracting platform for use across civilian federal agencies. Bloomberg has reported that the LLMs come with contract terms in place, which will enable and empower federal agencies to adopt the models and use them at rapid pace. The GSA have described the decision to approve Google, OpenAI and Anthropic as significant and serves as a clear indication of their efforts to responsibly adopt and scale transformative AI technologies. 'Making these solutions available through GSA's MAS will facilitate strong, widespread federal agency adoption and ensure easy access to AI tools that will improve their everyday workflows and processes.' In addition to this, the GSA highlighted how the decision is aligned with US President Donald Trump's AI action plan, which has been engineered to position the United States as the global leader when it comes to AI despite the stiff opposition coming in the form of fellow economic heavyweight China. The move by the GSA also supports Trump's war on government DEI programs, that the US President vowed to disband if re-elected to the White House, he has previously descried the DEI programs as immoral and illegal. 'As we procure these products, we're focused on models that prioritise truthfulness, accuracy, transparency, and freedom from ideological bias, aligning with the Trump administration's policy that federally procured AI systems must prioritise truth and accuracy over ideological agendas,' Federal Acquisition Service Commissioner Josh Gruenbaum stated. He noted through GSA's marketplace, 'agencies will be able to explore a wide range of AI solutions, from simple research assistants powered by large language models to highly tailored, mission-specific applications'. For Anthropic, Google and OpenAI is it a big win, as having their products on the GSA provides a trusted path into the federal marketplace. The financial terms and length of contracts were not available.
