Latest news with #Invoke
The Star
17 hours ago
- Politics
- The Star
Contradictheory: A buffet of personal data up for grabs?
It must have been about 25 years ago when I attended a talk about the power of integrating databases, and back then we were lucky governments hadn't quite figured it out yet. But obviously all that has changed now. The speaker was from Canada, and he illustrated the concept with an example: On one hand, the government struggled to catch welfare recipients who were quietly nipping across the border to work in the United States. On the other hand, it would have been trivially easy to catch them if someone just thought to cross-reference the welfare database with weekday border crossing records. The technology wasn't the issue. The stumbling block was getting different government departments to cooperate. But while this was an opportunity begging to be taken, the speaker also highlighted the potential dangers: The power inherent in data sharing, when coupled with the reach and authority of national governments, could lend itself to a wide scope of abuses. Which is precisely what is happening in the United States at the moment. Reports indicate that Elon Musk's Department of Government Efficiency (DOGE) has equipped agents with 'backpacks full of laptops, each with access to different agency systems... to combine databases currently maintained separately by multiple federal agencies'. Why are they doing this? Because DOGE wants to collate immigration data and cross-reference it with tax records, welfare claims, and even state voting records. According to critics, they allegedly want to target immigrants not just for tracking, but possibly for persuasion, prosecution, and persecution. Civil liberties groups are sounding the alarm, pointing out that this level of cross-agency integration is ripe for abuse. Truth is, this isn't something new or foreign for Malaysia. Back in 2018, politician Rafizi Ramli openly discussed how his startup, Invoke, was leveraging big data to support the then Opposition candidates. Using electoral roll data, phone polls, and surveys, they constructed detailed voter profiles based on age, postcode, gender, race, and religion to create a 'microtargeting' strategy to identify and win over swing voters with a 40%-60% chance of flipping. Fast-forward to today, and the government's interest in big data is no secret. From the MyDigital Blueprint to the freshly minted Data Sharing Act 2025, the rhetoric is all about 'digital transformation' and 'evidence-based policymaking'. Then in April, the Malaysian Communications and Multimedia Commission (MCMC) told telcos to hand over mobile network usage data, down to your call records, IP connections, and yes, your location coordinates. Failure to comply could result in a fine of RM20,000 or up to six months in official line is that this is for the 'generation of official statistics to support evidence-based policymaking' in the ICT and tourism sector. (Specifically, to identify the number of mobile broadband subscriptions, and to track the number of visitors and 'domestic tourist trips'.) The government assures us that all data handed over is anonymised and contains no personally identifiable information. But I've learned to be sceptical following alleged breaches that reportedly had data from millions of Malaysians in government databases being sold online. Now we are to trust the authorities when they say they want the data for policymaking? It's a bit of an overkill, like queuing up for an hour for a free plate of nasi kandar. (Which I have done before.) But why settle for a plate when you could have the whole buffet of personal information data? A tech news website said what many of us might be thinking: 'What just happened doesn't feel like planning. It feels like surveillance.' Local technologist and computer scientist Dinesh Nair was more colourful, labelling MCMC's explanation as (if I may paraphrase) horse excrement, noting that location data coupled with the time could be used to positively identify individuals. This isn't theory. We've known for decades how it's possible to do 'data re-identification'. In 1997, MIT computer scientist Latanya Sweeney famously cross-referenced public voter records with anonymised health data to identify then-governor of Massa-chusetts, Bill Weld. She even mailed him his own medical records as proof. And a few years later, I was reviewing drafts of what would eventually become Malaysia's Personal Data Protection Act (PDPA). While the PDPA was a good step forward in defining personal data rights – like requiring consent and allowing individuals to revoke it – it had one gaping hole: it doesn't apply to government entities. Section 3(1) of the Act explicitly exempts federal and state bodies. I was told at the time that government departments handled so much data, it would bog them down in red tape to comply. That was the official line. I suspect the unofficial thinking has evolved into: 'Why would we give up control of such a rich treasure trove?' Fixing this isn't hard. Step one: remove the PDPA exemption for government agencies. Just cross out a paragraph. Step two: appoint an independent oversight body where people can lodge complaints if their rights are violated. The latter's nearly in place, given that the current government has promised an Ombudsman Bill this year that could serve this very function. At the very least, amending the Act would allow us to point to a breach and say, 'That's illegal', instead of shrugging and saying, 'Well, it's the government. What can you do?' Because we were warned. We just didn't listen. In his fortnightly column Contradictheory, mathematician-turned-scriptwriter Dzof Azmi explores the theory that logic is the antithesis of emotion but people need both to make sense of life's vagaries and contradictions. Write to Dzof at lifestyle@ The views expressed here are entirely the writer's own.
Yahoo
11-02-2025
- Entertainment
- Yahoo
This Company Got a Copyright for an Image Made Entirely With AI. Here's How
The woman has three eyes, spaghetti noodle-like hair and a slice of yellow, melted cheese dripping down her head. Her body is a mosaic, with primary-colored shards of glass similar to something you might see in a church's stained glass window. I've seen a lot of AI images in my time as an AI reporter, and this one certainly has that je ne sais quoi AI look about it. The image is called A Single Piece of American Cheese, and it was conjured up by Kent Keirsey, CEO of Invoke, an AI creation platform. While it seems like just another AI image at first glance, it's significant for another reason. The image was created entirely with generative AI -- and Invoke just got a copyright for it. If you're surprised by that, you're not alone. I was too. Here's why: This new era of AI-generated images has stirred significant concerns about creativity and originality and the role of humans in the creation of art. It's also become a battleground over ownership and protecting the rights of the artists and illustrators who make the art that then gets used by the companies behind AI image generators. I talked with Keirsey and others involved in A Single Piece of American Cheese and the copyright award to understand how it all came together. It's a fascinating look behind the scenes at the questions and conundrums of art and AI. Keirsey first generated an AI image with Invoke. Then, he used a process called "inpainting," which allowed him to highlight specific regions of the image and generate new AI elements in that area with a new prompt. He added roughly 35 of these AI edits to the original AI image, ultimately resulting in the final image. Keirsey's hands-on role in the creation process was key to his application for copyright. Invoke first applied for a copyright for A Single Piece of American Cheese in August 2024 and was denied because the US Copyright Office said it "lack[ed] the human authorship necessary to support a copyright claim," according to correspondence between Invoke's legal team and the office reviewed by CNET. Invoke followed up with more evidence, including a timelapse video of the image's creation and an explanation of how Keirsey was involved in the creation process. The Copyright Office gave its stamp of approval on Jan. 30. The office granted approval and said it determined the image "contains a sufficient amount of human original authorship in the selection, arrangement, and coordination of the AI-generated material that may be regarded as copyrightable." The key words in there are "selection," "coordination" and "arrangement." In the certificate of registration, also viewed by CNET, the office said that the AI-generated components were excluded from the copyright claim. It was Keirsey's collaging of all these elements that resulted in creating something new enough, something expressing sufficient human creativity, that was eligible for copyright. And in this case, the Copyright Office granted the claim. The Copyright Office has been navigating the legal and ethical complexities of protection in the age of AI since it launched its AI initiative in March 2023. Since then, hundreds of creators have registered and received copyrights for works containing AI. Pursuant to guidance from the federal agency, creators have to disclose any AI usage in their work. Those AI elements are considered pre-existing material and aren't covered by the copyright. Instead, it's the new things people are adding -- their original imagery, text and so on -- that are covered. In Keirsey's case, the composite image, made by his deliberate manipulation of each piece, is what's protected. In other words, the individual parts of the image aren't copyrighted, but the sum of those parts is, thanks to Keirsey's direction. "I give the example of making a collage out of old photos. So there, imagine you find a box of photos that are 100 years old. They're out of copyright, nobody owns them, and then you make a two-dimensional collage out of these old photos. [That] collage that would probably be protectable under the Copyright Act, because even though the individual photos aren't protected, what you've done with them is creative and warrants protection," said Judd Lauter, special counsel at Cooley LLP, who worked on the copyright application for Invoke, in an interview. This logic falls in line with the latest guidance the Copyright Office, published just last month, on the copyrightability of AI images. The report said that images edited or enhanced with generative AI could be eligible for copyright, like movies that use AI de-aging tech or photos retouched with generative editing. The prompts that create AI images aren't copyrightable since it's the generators that interpret and create the images, potentially limitlessly. The Copyright Office maintained in that guidance that images entirely created by AI without sufficient human editing, however, still can't be copyrighted since there's not enough human contribution (or authorship) in the process. So if you typed a prompt into Dall-E 3 and copied and pasted whatever image popped up into an application for copyright, you would likely be denied. What's interesting is that every element of A Single Piece of American Cheese is AI-generated -- there isn't a human-generated base photograph or design underneath it all. In a statement to CNET, the Copyright Office said its policy is not to comment on specific registrations or claimants. The image has a kind of "absurdity," as Keirsey calls it, that he said was inspired by and reflects his experience trying to get copyrights on AI images. "The woman is made of this kind of fractured stained glass, and that is, in essence, what we got copyright on. We did not get copyright on each individual shard, or each individual element, because that's AI-generated. But what we did get a copyright on is the whole of that composition -- the shards of glass, if [they] were melted together by American cheese," said Keirsey in an interview. Human authorship, meaning the work in question was created by a real-life human, is one of the most fundamental requirements for copyright. But in the age of AI art, there are a lot of debates and questions about just how involved actual humans are in the creative process. Keirsey and his legal team argued that Invoke's inpainting tool and other features gave him enough creative control over the work to warrant a copyright. It's important to note that the very basis of AI generative content relies on training based on existing content, primarily created by people. Over the past few years, content creators, publishers and artists have been extremely concerned about how their work is used in this training process. Groups like The New York Times and bands of artists are in the midst of heated legal battles over the legality of such usage. Other folks, like content catalog and media owners, have negotiated with AI companies to license their content for training in multimillion-dollar deals. Invoke uses open-source foundational models (including Stable Diffusion 1.5), doesn't train on its users' work and its users own the work they create on the platform. Invoke's success in copyright protection certainly offers hope for creators looking to copyright their AI-edited work, but there's no guarantee that every piece will meet the threshold of sufficient human authorship, as deemed the Copyright Office. What is significant about this news is that it establishes a guidepost for other creators, Lauter said. "What we've demonstrated here is that we've defined a threshold. I think creators are going to continue to test what those boundaries are, both in terms of how human creativity is applied and embodied within content created with the assistance of AI, and also how much creativity is going to be needed to reach that threshold. And the way we figure it out is through example," said Lauter. While Keirsey has no immediate plans for A Single Piece of American Cheese now that it's copyrighted, he sees it as an important milestone. The legal, ethical and bigger philosophical questions around AI images and copyright will continue. But there's still a place for human creativity with AI, Keirsey said. "I think our relationship with technology is changing, and it's going to be different," said Keirsey. "It's important for people to realize that humans aren't going away. Just as with all technology, we're finding a new way to relate to it, and the creativity that we have matters just as much as it did five years ago, before we had all of this stuff."
