Artificial intelligence techniques between image documentation, energy consumption and its governance

At a time when artificial intelligence technologies are capable of producing hyper-realistic images and modifying photographs in such a way that it is difficult to distinguish between the real and the artificial, a new issue arises that goes beyond the quality of the image itself: How can we prove that what we see in a photo actually happened? It seems that Apple is trying to answer this question through a new feature that it added to the iPhone 18 Pro, which allows the creation of a copy of the image whose originality and source can be verified.

The camera of the new iPhone 18 Pro from Apple includes a low-profile feature aimed at addressing one of the biggest problems associated with artificial intelligence, according to the Independent newspaper.

Apple unveiled the new phone last week during an event in which the company focused heavily on the camera as one of the device's main features. Apple highlighted a new feature of variable aperture, which allows the camera to work much more efficiently in low-light conditions, in addition to a set of new software tools that give users greater control over images.

However, there is another feature that received only a passing mention during the event, although it may represent a more significant change in the way photos are shared online. This feature is called “Apple Reference Image,” and it aims to enable photographers to create an unedited, verifiable version of the photos they take with the iPhone camera. Thus, the photographer can provide evidence that the image documents a real scene that he or she actually saw and captured, rather than being an image generated or substantially modified using artificial intelligence tools.

This step targets a wide-ranging problem that is becoming more complex with the rapid development of artificial intelligence tools, which is the difficulty of proving that the image is actually real, and that it reflects reality as it is supposed to do. As the capabilities of these tools have increased and spread widely, distrust has grown towards visual content of all kinds, including photographs, due to the ease with which they can be forged or modified in ways that appear very convincing.

This problem is not limited to external tools designed to generate or modify images; Apple itself already provides some tools that allow for image modifications. For example, its latest phones include a “Clean Up” tool, which allows users to remove distracting elements from the background of an image.

The issue becomes more complex with the great development witnessed in image processing technologies integrated into smartphones. Many phone cameras today rely on highly sophisticated processing and editing tools, which in some cases makes the final image significantly different from the information originally captured by the camera sensor.

For example, in 2023, Samsung phone users discovered that AI-powered processing tools, which are designed to improve the appearance of photos, can replace some original elements with completely artificial ones. When the user tries, for example, to take a picture of the moon, the phone can recognize what he is trying to photograph, and then replace the image taken by the sensor with an artificial, clearer version of the moon, which raises questions about the extent to which the final image reflects the scene that the user actually captured.

In a white paper published on Apple Reference Image, Apple acknowledged that the easy availability of AI tools allowed it to provide “useful features,” including tools found in its own phones. But at the same time, she pointed out that these same tools “make it difficult to distinguish between images documenting real events and artificial images that have been substantially modified or completely created.”

The company explains that “this problem is not easy to solve,” noting that modern cameras include a set of “advanced image processing algorithms to produce the final viewable image.” Therefore, any system that aims to verify the authenticity of images must not only examine the final image, but must also be able to verify the software used to process it, along with the data that actually arrived from the camera sensor.

From this standpoint, “Apple Reference Photo” seeks to provide a way to help separate the photo that was actually taken by the camera from the photo that later underwent major modifications or changes.

There is a paradox that is growing as the world becomes more dependent on artificial intelligence: the algorithms that we expect to make our lives, our cities, and our industries more efficient, themselves require huge data centers, increasing amounts of electricity and water, and infrastructure whose effects extend beyond the digital world.

But what if artificial intelligence, which consumes energy, could simultaneously help us save it?

This question emerged strongly at the Fourth UNESCO Global Forum on Artificial Intelligence Ethics, held at the Ritz-Carlton Hotel in Riyadh from September 15 to 17, 2026, where the discussion expanded from the traditional principles of responsible artificial intelligence to a more practical question: How can these principles be transformed into applicable and measurable decisions and tools?

During the forum, UNESCO launched new tools, including the advanced version of the “Readiness Assessment Methodology” (RAM 2.0), and an artificial intelligence and environment tool aimed at helping policymakers reduce the environmental footprint of artificial intelligence systems, while at the same time investing in their capabilities in climate action, protecting biodiversity, and enhancing the resilience of ecosystems. According to UNESCO, 58 countries have already completed readiness assessments using its methodology.

A two-way relationship

In a session that dealt with “Artificial Intelligence, Environment, Environmental Systems, and Energy,” Yahya Khoja, Director General of Artificial Intelligence and Business Development at the Saudi Ministry of Energy, presented a remarkable equation for this relationship.

He explained, citing what the Saudi Minister of Energy affirmed, that the relationship between energy and artificial intelligence goes in two directions: artificial intelligence needs energy to continue its rapid growth, while the energy sector, in the opposite direction, can benefit from artificial intelligence models and solutions to raise efficiency and address its challenges.

Here, according to Khoja, the “balance” challenge arises for governments: How can we support the growth of technology and data centers, while at the same time using resources efficiently, and taking into account sustainability, environmental and economic considerations?

Khoja presented another dimension related to the role that Saudi Arabia can play as a “solutions country,” capable of developing experiences and practices that can be shared with the world.

Artificial Intelligence... Energy Inspector

The conversation did not remain at the level of concepts. Khoja presented an example of using data and smart analytics to understand electricity consumption and energy efficiency in government assets and facilities, from educational and health facilities to street lighting.

The challenge lies in scale; When huge numbers of assets are spread over a wide geographic area, sending human teams to inspect each facility and identify inefficiencies and waste becomes an extremely complex task. Here, artificial intelligence can analyze huge amounts of data, monitor abnormal patterns in consumption, and help identify locations that deserve inspection, maintenance, or efficiency improvement, allowing human intervention to be directed to where the need is greatest.

Thus the paradox becomes clear: Artificial intelligence is not only a consumer of energy; Rather, it may become a tool for discovering its waste.

Science confirms the paradox

This dual relationship finds support in recent research. In a scientific review published in the journal Nature Reviews Clean Technology in July 2026, researchers warned that the accelerating expansion of data centers supporting artificial intelligence is putting increasing pressure on electricity and water grids and emissions, and some estimates reviewed by the study indicate that the demand associated with data centers may rise sharply until 2028, with scenarios reaching about 650 terawatt-hours per year for artificial intelligence alone.

But the other side of the paradox appears in a study published in 2026 on using artificial intelligence to improve energy efficiency in buildings. Using consumption prediction and dynamic adaptation to operating conditions, the average daily consumption in the experimental group decreased from 218.4 to 174.5 kWh compared to the conventional control system. This result cannot be generalized to all buildings, but it provides a practical example of the ability of artificial intelligence to manage energy, not just its consumption.

The answer is not technical

But perhaps Hoxha's most telling remarks came when he was asked what advice he would give to a government trying to align AI and energy policies.

His answer was not a new algorithm or a more advanced data center. Rather, he said: “I do not think the answer is technical... The answer is actually dialogue.” He explained that what is required is to bring together stakeholders to discuss difficult questions; It cannot be decided in one session or within seconds, but rather requires listening to the different parties and taking action collectively.

And here we may reach the essence of artificial intelligence ethics in its new phase: governance is not achieved by writing principles alone, and sustainability is not achieved by simply describing the technology as “green”. Rather, artificial intelligence, energy, environment, economics, and regulation should all sit at the same table.

From principles to implementation

Since the adoption of the UNESCO Recommendation for the Ethics of Artificial Intelligence in 2021, the global question has begun to shift from “What principles do we want?” to “How do we apply it and measure its results?” At the Riyadh Forum, this transition appears clear in the quest to transform ethical governance from general principles into applicable tools and decisions.

In the energy sector, the test becomes tangible: a data center, a power grid, a government building or a hospital. Artificial intelligence needs energy to advance, and energy can be used by artificial intelligence to become more efficient.

As for what links the two to a sustainable future, it may not be another algorithm... but rather governance that knows how to bring together technology, energy, and people around the same decision.