#123 The Reconnaissance Revolution and its Discontents
In this edition of Technopolitik, Aditya Ramanathan talks about the reconnaisance satellite revolution that was once said to ensure strategic stability. Anwesha Sen follows with a piece on the recent controversy surrounding xAI’s chatbot Grok.
This newsletter is curated by Adya Madhavan.
Antarikshmatters: Satellites in the Crosshairs
— Aditya Ramanathan
In 1986, the American historian John Lewis Gaddis published a paper titled The Long Peace. Writing in the waning days of the Cold War, Gaddis turned his attention to the four prior decades - and asked a big question: what explained the absence of major war between the superpowers? In searching for answers, Gaddis doffed his hat at the usual suspects like bipolarity in the international system, and nuclear weapons.
Then he turned his attention to what he called the ‘reconnaissance revolution’.
Here, Gaddis argued that overflight by satellites had afforded the superpowers much greater transparency and insight into each other's military movements than had ever been the case before. As he put it:
“The effect has been to give each side a far more accurate view of the other's military capabilities-and, to some degree, economic capabilities as well-than could have been provided by an entire phalanx of the best spies in the long history of espionage. The resulting intelligence does not rule out altogether the possibility of surprise attack, but it does render it far less likely, at least as far as the superpowers are concerned.”
Gaddis also acknowledged the role satellite reconnaissance had played in nuclear arms control. Spies in orbit had allowed both superpowers to “trust but verify”, as Ronald Raegan’s favourite Russian proverb went. In short, Gaddis argued, reconnaissance satellites had made military surprise harder to achieve and arms control agreements easier to sustain.
Breaching the Sanctuary
What Gaddis left unexamined was whether the reconnaissance revolution would hold once shots were fired in anger. In 1986, there was good reason to think low Earth orbit (LEO) assets were vulnerable to attack from a variety of antisatellite (ASAT) weapons that both the US and USSR had tested over years. There were also far fewer satellites in orbit in 1986 than today, meaning the debris from kinetic strikes was a manageable problem. More importantly, satellites were expensive, large in size, and few in number. This meant ASATs enjoyed the benefits of a small target set and favourable exchange ratios.
In the years since Gaddis’ paper, one might have expected spacefaring states would adapt to the realities of ASATs. But the relaxation of great power tensions in the first twenty years after 1986 meant there was little urgency.
That changed around 2007 when China conducted a destructive direct ascent (DA) ASAT missile test. This triggered a series of destructive tests from the US, India and Russia. Since 2007, we’ve also seen the development of other counterspace capabilities, from dazzling lasers to rendezvous and proximity operations (RPOs).
The most significant response to these counterspace capabilities has been the idea of proliferated constellations - large numbers of inexpensive and replaceable satellites. Some constellations predate this idea. For instance, global navigation satellite systems (GNSS) consist of 24-45 satellites, though these are neither cheap nor terribly expendable. However, proliferated constellations are most epitomized by satellite internet systems such as Starlink as well as projects that envision hundreds of reconnaissance satellites.
It’s possible to argue proliferated constellations are already working. At least one of the reasons why Russia has not attacked Starlink satellites (which are providing vital communications support to the Ukrainian military), is that tools like DA-ASATs are impractical against hundreds or thousands of orbital craft.
Proliferation May Not Equal Protection
However, it may be reasonable to expect a new round of countermeasures against proliferated constellations. In a 2024 paper, Ivan Oelrich, Paul van Hooft and Stephen Biddle argued that states can develop effective countermeasures against proliferated reconnaissance constellations.
One, they pointed out that DA-ASAT missiles still enjoy a favourable exchange ratio against satellites, even most small, cheap ones. Two, they acknowledge that most effective countermeasures would be terrestrial and non-kinetic, primarily involving dazzling and jamming.
Let’s start with dazzling. Terrestrial lasers of this sort already exist in limited numbers. The authors argue that relatively low-powered lasers can perform the job of temporarily disabling lenses and other sensors on adversary satellites. While the satellites will otherwise remain functional, they can be stopped from performing some key tasks.
Of course, such lasers are likely to be met with a response from spacefaring states. As the authors argue:
“Once ASAT lasers are deployed, military satellite designers will incorporate protection that will seriously degrade laser effectiveness, such as hardening, shielding, concealment, and deception. For example, if satellites know the location of fixed lasers (as they will once the laser is used), then satellites could turn vulnerable solar panels edge-on toward the laser. Satellite bodies could be protected with highly reflective, heat tolerant material. Sensors and antennas might need to be covered or turned away, blinding or silencing the satellite”.
The downside from these measures is that they add to the weight, complexity and cost of satellites. Another option for spacefarers is to simply increase the number of satellites surveilling an area at any particular time, thus overwhelming terrestrial dazzlers. However, given that these satellites are likely to be in LEO, such constellations are likely to require thousands of orbiters.
The authors also consider synthetic aperture radar (SAR) satellites. These would require jamming rather than dazzling. The authors calculate that a truck carrying a jammer with a power of just a few hundred watts, can effectively jam an SAR satellite. Since the sources of jamming are easier to identify, the jammers would have to shoot-and-scoot, turning off their devices and moving periodically, perhaps handing over the task to another truck-mounted jammer..
While the authors restrict themselves to considering reconnaissance satellites and don’t claim lasers and jammers will render satellites useless, their paper does suggest that militaries have an incentive to invest in existing technology to develop effective non-kinetic ASAT solutions. The effect of this is likely to be less transparent battlefields as well as greater contestation in peacetime, as states seek to prevent each other from gaining useful pattern-of-life information. The reconnaissance revolution has always had its discontents. They may soon have reason to cheer.
Technopolitik: The Grok Controversy: Legal Implications and Debates on AI Liability
— Anwesha Sen
The recent furore surrounding Grok, a generative AI chatbot integrated into X (formerly known as Twitter), has brought back significant discussions regarding the legal landscape for AI chatbots in intermediary platforms, particularly in relation to the safe harbour provision, liability issues, and the application of Section 67 of the Information Technology (IT) Act in India to the forefront.
Grok is designed to interact with users, answer questions, and provide helpful recommendations, all within the platform. While it aims to enhance user experience, its integration has sparked considerable controversy due to concerns over its responses and the potential spread of misinformation. Users and legal bodies have raised alarms about the chatbot's ability to generate content in profane language, which has led to proposals for legal action against the platform.
At the heart of the debate is the issue of safe harbour, a legal provision that protects platforms from liability for content generated by users or, in this case, by AI systems like Grok. Under Section 79 of the IT Act in India, platforms are generally shielded from legal responsibility for user-generated content, provided they comply with due diligence requirements. However, this protection does not apply when the platform itself is directly involved in the creation or dissemination of illegal content. As AI chatbots like Grok become more sophisticated, questions arise about whether these platforms should continue to be shielded from liability, especially when they provide a medium for potentially harmful AI-generated content.
The challenge, then, is determining whether Grok’s interactions fall under the safe harbour provision or whether it opens up a new category of liability for X and other platforms that use similar technologies. Advocates for stronger regulations argue that as AI chatbots become more pervasive and capable, they should be held accountable for the content they generate, particularly when it leads to harm or misinformation.
In India, Section 67 of the IT Act, which deals with the publishing of obscene content, is also relevant to this discussion. Section 67 outlines punishment for publishing or transmitting obscene and/or sexually inappropriate content. Obscene content is defined as content that is likely to corrupt or harm those who see, read, or hear it. Interpretation of this definition is subjective and it is unclear whether the use of profanity (which is not uncommon on social media platforms) would fall under obscene content. The question then arises whether platforms like X could be held liable under this section.
As Grok and similar technologies evolve, the need for clearer legal frameworks around AI, liability, and platform responsibility becomes ever more pressing. The outcome of the legal discussions surrounding Grok will likely set important precedents for AI-driven platforms globally.
*ChatGPT has been used to refine the language of this piece.