Google’s AI research lab DeepMind has announced that in partnership with Google Cloud, it’s launching a beta version of SynthID, a tool for watermarking and identifying AI-generated images.

The AI Image Challenge

Generative AI technologies are rapidly evolving, and AI-generated imagery, also known as ‘synthetic imagery,’ is becoming much harder to distinguish from images not created by an AI system. Many AI generated images are now so good that they can easily fool people, and there are now so many (often free) AI image generators around and being widely used that misuse is becoming more common.

This raises a host of ethical, legal, economic, technological, and psychological concerns ranging from the proliferation of deepfakes that can be used for misinformation and identity theft, to legal ambiguities around intellectual property rights for AI-generated content. Also, there’s potential for job displacement in creative fields as well as the risk of perpetuating social and algorithmic biases. The technology also poses challenges to our perception of reality and could erode public trust in digital media. Although the synthetic imagery challenge calls for a multi-disciplinary approach to tackle it, many believe a system such as ‘watermarking’ may help in terms of issues like ownership, misuse, and accountability.

What Is Watermarking?  

Creating a special kind of watermark for images to identify them as being AI-produced is a relatively new idea, but adding visible watermarks to images is a method that’s been used for many years (to show copyright and ownership) on sites including Getty Images, Shutterstock, iStock Photo, Adobe Stock and many more. Watermarks are designs that can be layered on images to identify them.  Images can have visible or invisible, reversable, or irreversible watermarks added to them. Adding a watermark can make it more difficult for an image to be copied and used without permission.

What’s The Challenge With AI Image Watermarking? 

AI-generated images can be produced on-the-fly and customised and can be very complex, making it challenging to apply a one-size-fits-all watermarking technique. Also, AI can generate a large number of images in a short period of time, making traditional watermarking impractical, plus simply adding visible watermarks to areas of an image (e.g. the extremities) means it could be cropped and the images can be edited to remove it.

Google’s SynthID Watermarking 

Google SynthID tool  works with Google Cloud’s ‘Imagen’ text-to-image diffusion model (AI text to image generator) and uses a combined approach of being able to add and detect watermarks. For example, the SynthID watermarking tool can add an imperceptible watermark to synthetic images produced by Imagen, doesn’t compromise image quality, and allows the watermark to remain detectable, even after modifications (e.g. the addition of filters, changing colours, and saving with various lossy compression schemes – most commonly used for JPEGs). SynthID can also be used to scan an image for its digital watermark and can assess the likelihood of an image being created by Imagen and provides the user with three confidence levels for interpreting the results.

Based On Metadata 

Adding Metadata to an image file (e.g. who created it and when), plus adding digital signatures to that metadata can show if an image has been changed. Where metadata information is intact, users can easily identify an image, but metadata can be manually removed when files are edited.

Google says the SynthID watermark is embedded in the pixels of an image and is compatible with other image identification approaches that are based on metadata and, most importantly, the watermark remains detectable even when metadata is lost.

Other Advantages 

Some of the other advantages of the SynthID watermark addition and detection tool are:

– Images are modified so as to be imperceptible to the human eye.

– Even if an image has been heavily edited and the colour, contrast and size changed, the DeepMind technology behind the tool will still be able to tell if an imaged is AI-generated.

Part Of The Voluntary Commitment

The idea of watermarking to expose and filter AI-generated images falls within the commitment of seven leading AI companies (Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI) who recently committed to developing AI safeguards. Part of the commitments under the ‘Earning the Public’s Trust’ heading was to develop robust technical mechanisms to ensure that users know when content is AI generated, such as a watermarking system, thereby enabling creativity AI while reducing the dangers of fraud and deception.

What Does This Mean For Your Business?

It’s now very easy for people to generate AI images with any of the AI image generating tools available, with many of these images able to fool the viewer possibly resulting in ethical, legal, economic, political, technological, and psychological consequences. Having a system that can reliably identify AI-generated images (even if they’ve been heavily edited) is therefore of value to businesses, citizens, and governments.

Although Google admits its SynthID system is still experimental and not foolproof, it at least means something fairly reliable will be available soon at a time when AI seems to be running ahead of regulation and protection. One challenge, however, is that although there is a general commitment by the big tech companies to watermarking, the SynthID tool is heavily linked to Google’s DeepMind, Cloud and Imagen and other companies may also be pursuing different methods. I.e. there may be a lack of standardisation.

That said, it’s a timely development and it remains to be seen how successful it can be and how watermarking and/or other methods develop going forward.

Following several months of testing, Firefox users can now take advantage of the Firefox Relay email masking tool from within the browser to help preserve their online anonymity and boost security.

What Is Firefox Relay? 

Firefox Relay from Mozilla is a Firefox browser extension that lets users keep their email address private when filling out online forms. It does this by creating an email mask (a forwarding email address that’s different to their real email address) and forwards messages to the user’s real email address to keep it hidden. Users can disable or delete the mask when it’s no longer needed.

Why Mask Your Email? 

There are number of good reasons why users may want to mask (hide) their real email address, including:

– Feeling safer when dealing with companies they don’t trust.

– Preventing spam.

– Protecting your real email address from data breaches.

– Protecting your real email address from being sold or shared.

– Email masks can be easily deactivated with no consequences (unlike a user’s real email account).

How Does Firefox Relay Work?

Users just need a Firefox Account and the (free) Firefox Relay browser extension. To activate it, it’s a case of clicking on the Firefox Relay icon in the toolbar, following the prompts to sign in/up to Firefox Accounts, and registering the email address to forward emails to.

To enable/disable Relay, users need to click on the icon in the toolbar, go to Settings, and set the toggle button for “Show Relay icon on email fields on websites” to on or off (green means it’s on).

Once switched to on, when users visit an online form, the Firefox Relay icon appears in the email field, and it’s just a case of clicking it to generate an email mask using a name of the user’s choosing or to see a list of recently used masks.

Challenge 

One challenge for Firefox Relay is that some services don’t allow users to sign up using email masks. Last year, for example, popular cloud-based development platform GitHub blacklisted Firefox Relay domains.

Other Privacy Tools Available 

Firefox, however, is in low fourth place within the Browser market with less than 3 per cent market share (Chrome has over 60 per cent, Safari 24, and Edge 5) so its reach as a privacy tool is slim. Also, Relay has plenty of competition when it comes to the wider market of privacy tools, e.g. VPNs and (other) secure browsers like Brave, DuckDuckGo, Tor, and even Chrome. Furthermore, Relay has competitors in the shape of temporary/disposable email platforms like Tempmail, EmailOnDeck, DispoasbleMail (and more), as well as having more direct email masking competitors such as FastMail and its email-masking browser extension IronVest.

What Does This Mean For Your Business? 

With email still such a vital business tool and with the huge increases in phishing attacks and news of more major data breaches in recent times, plus with tighter data protection regulations to comply with, it’s no wonder email masking may be appealing to many users.

As Mozilla points out, email masking can also be a way of cutting down on the amount of spam that blocks business email accounts, causing frustration, wasting valuable business time, and obscuring opportunities. Email masking, such as that provided by Relay may also make users feel safer, feel more in control, help increase their general security and reduce risk, e.g. when contacting companies they may not fully trust.

Firefox now has a relatively low browser market share and the advantages of giving users a fast, convenient, and easy way to generate and switch off masking email addresses straight from a trusted browser (with an extension rather than downloading yet another unknown app) may be a way to tempt new and returning users to Firefox to give it a try, thereby increasing Firefox’s competitiveness and share.

The recently launched Pyxis Ocean cargo ship uses giant foldaway ‘WindWings’ to help supplement engine power and could cut lifetime carbon emissions by 30 per cent.

The Cargo Ship 

The A Mitsubishi Corporation cargo ship, is currently demonstrating the WindWings, developed by BAR Technologies and Yara Marine Technologies, during its six week-long maiden voyage from China to Brazil. The ship, chartered by Cargill Ocean Transportation, left from Singapore on 25 August carrying 81,000 tonnes of cargo.

The WindWings 

The extraordinary features of the Pyxis Ocean cargo ship are its two rigid, foldaway 37.5-metre-tall steel and fibreglass ‘sails’ (WindWings) which automatically pivot to catch the wind. The WindWings, each made up of a centrally pivoted 10-metre-wide section with two five-metre-wide wings on either side can be folded down onto the cargo ship’s deck for arrival at ports and for passing through bridges or canals.

BAR Technologies estimates that the wind power harnessed by the WindWings, which can be retrofitted to many cargo ships (but not container ships), could deliver an average fuel saving of up to 30 per cent (on new build ships) and could save as much 1.5 tonnes of fuel per day on an average global route. Adding four wings on cargo could, therefore, save a massive 6 tonnes of fuel per day, stopping 20 tonnes of CO2 being produced.  BAR Technologies says the WindWings make the savings by combining “wind propulsion with route optimisation” and are initially aimed at being fitted to “bulk carriers and tankers.” 

Targets 

With more than 90 per cent of world trade being carried across oceans by 90,000 fossil-fuel powered marine vessels, the global freight shipping industry is a major producer on CO2 emissions, which contribute to climate change (and acidification). For example, it’s been estimated that the global shipping industry is responsible for more than 3 per cent of global CO2 emissions and, if it were a country, it would be the sixth largest producer of greenhouse gas (GHG) emissions.

For these reasons, the International Maritime Organisation (IMO) set the goal for the shipping industry ultimately arriving at a 50 per cent reduction in GHG emissions from 2008 levels (CO2, sulphur oxide and other gasses) by 2050.

Adding WindWings to cargo ships could, therefore, be one relatively fast way, in the absence of zero-carbon fuel for ships or a clear decarbonisation strategy, to at least begin reducing GHG emissions.

Innovation 

John Cooper, Chief Executive Officer, BAR Technologies said of the WindWings: “If international shipping is to achieve its ambition of reducing CO2 emissions, then innovation must come to the fore. Wind is a near marginal cost-free fuel and the opportunity for reducing emissions, alongside significant efficiency gains in vessel operating costs, is substantial.” 

What Does This Mean For Your Organisation?

When the alarm bells really started ringing back in 2008 about how the global shipping industry was producing more CO2 than most countries, twice as much as the air transport industry, and more than 3 per cent of global emissions, the ambitious but necessary 50 per cent cut by 2050 target was set. However, with the world still reliant on cargo ships for 90 per cent of international trade, no zero-emissions fuel on the horizon, a perceived relative inactivity among freight operators in trying to meet the target, and no clear strategy, the WindWings idea looks like a realistic option to get started. Although there’s ‘no one fits all’ and it may not work for container ships, they have been designed to be workable on the most common vessels sizes (using either 3 or 4 WindWings). The fact that they’re fully automated (touch of a button), can be folded away, have a lower power consumption and yet have the capacity to make major fuel and CO2 emission savings mean that, as long as they are affordable and reliable, they could provide a practical, and sustainable way forward for the freight shipping industry until other options are available. There’s also a kind of irony to the idea that the leading technology of today for greener shipping takes ships back full circle to the days of sails, with these just being more high-tech versions – an old solution to a newer problem.

28 Years Ago : eBay (& Amazon by Comparison)

Whilst eBay’s market cap is dwarfed by that of Amazon (i.e. circa 24 billion dollars compared to circa 1.4 trillion dollars), it’s easy to forget eBay helped shape online purchasing too.

Before it was rebranded as eBay in 1997, the site was originally called AuctionWeb. Founded by Pierre Omidyar in 1995, it was an experiment to create an online venue for person-to-person auctions and lacked any heavy investment. In fact, the hosting company were just charging thirty dollars per month to for the entire website when he started up, until they said they’d no longer host it for that price due to the growth in traffic, thereby forcing him to monetise his website via his sellers, which worked.

With his degree in computer science behind him and having worked in a subsidiary company of Apple, he was also working on various other web projects. He’d also co-founded another company called Ink Development, a company initially focused on developing software for pen-based computing. The company transformed its direction and became an e-commerce platform named eShop, which was later acquired by Microsoft in 1996.

In short, Pierre was the right person in the right place at the right time. And so was Elon Musk (more on him later).

The Unexpected First Listing

The first item ever listed on AuctionWeb was a laser pointer. To be more precise, it was a broken laser pointer. This wasn’t a mistake or an oversight. Omidyar had bought it for his own use but found out it was faulty. Rather than discarding it, he decided it would make an interesting first listing on his experimental auction site. He listed it clearly mentioning it was non-functional. To his astonishment, the laser pointer garnered bids and finally sold for $14.83.

Intrigued by someone paying for a known broken item, Pierre out to the buyer who simply responded that he was a collector of broken laser pointers. This quirky initial transaction showed the potential of a vast, unpredictable, and diverse online marketplace.

The Beanie Baby Phenomenon

While the broken laser pointer was AuctionWeb’s first sale, it was the Beanie Baby craze in the mid-90s that truly skyrocketed the platform’s popularity. Beanie Babies, those small plush animals filled with plastic pellets, became a massive collector’s item. AuctionWeb became a primary marketplace for these avid collectors, providing them a platform to buy, sell, and trade these toys.

The Beanie Baby phenomenon showcased the strength of the platform in bringing together niche audiences globally and the success also signalled the beginning of a new era where the average person could become an entrepreneur from the comfort of their own home.

From AuctionWeb to eBay

Seeing the growing potential, Omidyar soon renamed AuctionWeb to “eBay,” which was short for Echo Bay, the name of Omidyar’s consulting firm. The rest is history! eBay went public on September 24, 1998.

On its first day of trading, the stock’s price of $53.50 soared well past the initial target of $18, emphasising the optimism investors held during the dot-com boom. Unlike many that fell by the wayside, eBay is now a business that operates in over 30 countries with a market cap peaking at 80.6 billion dollars.

Amazon and eBay – Different Approaches.

While eBay and Amazon both started as online marketplaces in the 1990s, they evolved with different business models and strategies that have influenced their trajectories. eBay started off as an online auction place, although plenty of people and businesses sell via their ‘Buy Now’ function as an online shop.

There are many possibilities as to why eBay hasn’t reached the same magnitude as Amazon and here are a few :

1. Business Model: eBay began as a peer-to-peer auction site, allowing individual sellers and buyers to negotiate prices. This gave eBay a unique identity but also meant slower and less predictable transactions. Amazon, on the other hand, started as a book retailer and then expanded its product range, focusing on selling new products at fixed prices.
2. Fulfilment and Logistics: Amazon invested heavily in fulfilment-centres and logistics, creating a vast and efficient infrastructure for storage, packing, and shipping. This allowed them to ensure rapid delivery, leading to services like Amazon Prime. eBay, on the other hand, relies on individual sellers to handle shipping and logistics, which can be more variable in terms of speed and reliability.
3. Private Label & Product Expansion: Amazon developed its own private-label products and expanded into diverse categories. They also encouraged third-party sellers to use their platform, ensuring a vast product range.
4. Ecosystem Development: Amazon diversified its business areas, venturing into hardware (Kindle, Echo), streaming (Amazon Prime Video), cloud services (AWS), and more. This diversification created multiple revenue streams and bolstered its market presence.
5. Trust and Reliability: Amazon’s emphasis on customer service and consistent delivery times built significant trust with customers. While eBay has made efforts to ensure product authenticity and seller reliability, the peer-to-peer model sometimes leads to inconsistencies in product quality and delivery.
6. Global Expansion Strategy: Both companies pursued international expansion, but Amazon’s aggressive strategy of setting up localized versions of its site, fulfilment centres, and tailored services for different countries gave it a strong global footprint.
7. Subscription Model: Amazon Prime, a subscription-based service, not only offers faster delivery but also includes streaming, exclusive deals, and other perks. This has fostered customer loyalty and increased purchase frequency.
8. Feedback System: While eBay’s feedback system was innovative and built trust in the early days, some argue that it’s become less effective over time due to potential biases and reluctance from buyers and sellers to leave negative feedback.
9. Acquisitions and Divestitures: While both companies made acquisitions, their strategies differed. Amazon’s acquisitions like Zappos, Whole Foods, and Twitch were integrated into its ecosystem. eBay, on the other hand, made some large acquisitions, such as Skype (see below), which were later divested as they didn’t align with eBay’s core focus.

What can be seen is that eBay seems to have plateaued insofar as the gross merchandise volume (GMV) (i.e. the total amount of ‘stuff’ sold via the platform) is concerned whereas Amazon’s GMV is steadily rising and so is the share of that GMV being sold by third party sellers (rather than Amazon directly) so doubtless they’re eating some of eBay’s lunch.

Perhaps the different share prices reflect the differing optimism because if there’s one thing that investors like, it’s growth.. One thing that’s clear … both companies did well over the pandemic.

While eBay didn’t reach the same dizzying heights of Amazon, it’s nevertheless a true rags-to-riches success-story that’s worth studying, including a couple of their better-known acquisitions. Even if the acquisitions were later sold, it’s interesting to try and appreciate the thinking behind the strategy and synergy.

Payment Provider : X Marks The Spot

The X Factor Elon Musk’s wealth originated from a critical acquisition made by eBay in 2002. In the late 1990s, Musk co-founded X.com, an online payment company. X.com would later become known as PayPal after a series of developments and a merger.

It was this very company, PayPal, that eBay acquired in 2002 for $1.5 billion in stock. At the time of the acquisition, Musk held 11.7% of PayPal shares, translating to roughly $165 million from the sale. Not too shabby for Mr Musk and it certainly helped springboard his wealth to be in the same league as that of Jeff Bezos from Amazon.

Communications Considerations : Skype

In late 2009, eBay finalized the sale of Skype for an impressive $2.75 billion. This strategic move allowed eBay to refocus on its core e-commerce operations, while the deal also highlighted Skype’s significant growth and potential in the telecommunications sector.

The Future?

Whilst Amazon seems intent on taking over the world by expanding relentlessly into eBay’s territory (and many others), eBay will likely remain a trusted corner-of-the-web for people to buy and sell goods for many years to come.

All of which started 28 years ago (this week), with a good idea and a broken laser-pointer.

If you’d like to backup your ChatGPT chats, it’s possible to export them by email from within ChatGPT. Here’s how:

– In ChatGPT, click on the three dots (bottom left).

– Click on ‘Settings & Beta.’

– Click on ‘Data Controls.

– Click on the ‘Export Data’ button, read the information, and click on ‘Confirm export’ (if you agree).

– The data will be sent to your registered email in a downloadable file and will include account details and conversations.

In this article, we look at why Zoom found itself as the subject of a backlash over an online update to its terms related to AI, what its response has been, plus what this says about how businesses feel about AI.

What Happened? 

Communications app Zoom updated its terms of service in March but following the change only being publicised on a popular forum in recent weeks, Zoom has faced criticism because many tech commentators have expressed alarm that the change appeared to go against its policy to not use customer-data to train AI.

The Update In Question 

The update to Section 10 of is terms of service, which Zoom says was to explain “how we use and who owns the various forms of content across our platform” gave Zoom “perpetual, worldwide, non-exclusive, royalty-free, sublicensable, and transferable license and all other rights” to use Customer Content, i.e. data, content, communications, messages, files, documents and more, for “machine learning, artificial intelligence, training, testing” (and other product development purposes).

The Reaction 

Following the details of the update being posted and discussed on the ‘Hacker News’ forum, there was a backlash against Zoom, with many commentators unhappy with the prospect of AI (e.g. generative AI chatbots, AI image generators and Zoom’s own AI models namely Zoom IQ) and more) being given access to what should be private Zoom calls and other communications.

What’s The Problem? 

There are several concerns that individuals, businesses and other organisations may have over their “Customer Content” being used to train AI. For example:

– Privacy Concerns – worries that personal or sensitive information in video calls could be used in ways the participants never intended.

– Potential security risks. For example, if Zoom stores video and audio data for AI training, it increases the chance of that data being exposed in a hack or breach. Also, it’s possible with generative AI models that private information could be revealed if a user of an AI chatbot asked the right questions.

– Ethical questions. This is because some users may simply not have given clear permission for their data to be used for AI training, raising issues of consent and fairness.

– Legal Issues. For example, depending on the country, using customer data in this manner might violate data protection laws like GDPR, which could get both the company and users into legal trouble. Also, Zoom users or admins for business accounts could click “OK” to the terms of service without fully realising what they’re agreeing, to and employees who use the business Zoom account may be unaware of the choice their employer has made on their behalf. It’s also been noted by some online commentators that Zoom’s terms of service still permit it to collect a lot of data without consent, e.g. what’s grouped under the term ‘Service Generated Data.’

Another Update Prompted 

The backlash, the criticism of Zoom and the doubtless fear of some users leaving the platform over this controversy appears to have prompted another update to the company’s terms of service which Zoom says was to “to reorganise Section 10 and make it easier to understand”. 

The second update was a sentence, in bold, added on the end of Section 10.2 saying: “Zoom does not use any of your audio, video, chat, screen sharing, attachments or other communications-like Customer Content (such as poll results, whiteboard and reactions) to train Zoom or third-party artificial intelligence models.” 

On the company’s blog, Chief Product Officer, Smita Hashim, re-iterated that: “Following feedback received regarding Zoom’s recently updated terms of service Zoom has updated our terms of service and the below blog post to make it clear that Zoom does not use any of your audio, video, chat, screen sharing, attachments, or other communications like customer content (such as poll results, whiteboard, and reactions) to train Zoom’s or third-party artificial intelligence models.” 

The Online Terms of Service Don’t Affect Large Paying Customers 

Smita Hashim explains in the blog post that the terms of service typically cover online customers, but “different contracts exist for customers that buy directly from us” such as “enterprises and customers in regulated verticals like education and healthcare.” Hashim states, therefore, that “updates to the online terms of service do not impact these customers.” 

What Zoom AI? 

Zoom has recently introduced two generative AI features to its platform – Zoom IQ Meeting Summary and Zoom IQ Team Chat Compose, available on free trial and offering automated meeting summaries and AI-powered chat composition.

To customers worried that these tools may be trained using ‘Customer Content’ Zoom says, “We inform you and your meeting participants when Zoom’s generative AI services are in use” and has specifically assured customers that Zoom does not use customer content (e.g. as poll results, whiteboard-content, or user-reactions) to train Zoom’s own (or third-party) AI models.

Criticism 

In 2020, Zoom faced criticism over only offering end-to-end encryption as a paid extra feature after saying paying users would have it anyway. Also, with Zoom being the company whose product enabled (and is all about) remote working, it was criticised after asking staff living within a “commutable distance” (i.e. 50 miles / 80km) of the company’s offices to come to the office twice a week when it was reported to have said (at one time) that all staff could work remotely indefinitely.

What Does This Mean For Your Business? 

This story shows how, at a time when data is now needed in vast quantities to train AI, a technology that’s growing at a frightening rate (and has been the subject of dire warnings about the threats it could cause), clear data protections in this area are lagging or are missing altogether.

Yes, there are data protection laws. Arguably however, with the lack of understanding of how AI models work and what they need, service terms may not give a clear picture of what’s being consented to (or not) when using AI. There’s a worry, therefore, that boundaries of data protection, privacy, security, ethics, legality, and other contraints may be overstepped without users knowing it in the rush for more data as clear regulation is left behind.

Zoom’s extra assurances may have gone some way toward calming the backlash down and assuring users, but the fact that there was such a backlash over the contents of an old update shows the level of confusion and mistrust around this relatively new technological development and how it could affect everyone.