UPSC Prelims Practice Questions — How did Google’s AI satellite images raise safety concerns?

Question 1

In its public statement withdrawing the AI image-generation capability from Google Earth in 2026, Google attributed the decision primarily to which one of the following?

Accepted Answer

Screenshots of generated imagery shared by users that appeared to violate the company's own content policies

Answer

A binding takedown direction issued to the company by the European Commission under the EU's artificial intelligence law

Answer

A finding that the underlying model had been trained on licensed commercial satellite data without payment of royalties

Answer

A demand from the United Nations Office for Outer Space Affairs to suspend synthetic rendering of geospatial data

Question 2

Which one of the following user groups did Google itself name as the leading legitimate beneficiary of the Google Earth AI image feature even while announcing its withdrawal?

Accepted Answer

Geospatial professionals

Answer

Secondary-school geography teachers

Answer

Licensed real-estate developers

Answer

National disaster-response agencies

Question 3

With reference to Google's 'Nano Banana' family of image-generation models, consider the following statements:
1. The model was introduced in August 2025 as Gemini 2.5 Flash Image.
2. Images generated or edited with it in the Gemini app carry both a visible watermark and an invisible SynthID watermark.
3. Nano Banana Pro, released later, is powered by Google's Gemini 3 model.
4. The model is made available to developers exclusively inside the Gemini app and is not offered through Google AI Studio or Vertex AI.
Which of the above is/are NOT correct?

Accepted Answer

4 only

Answer

1 and 3

Answer

2 only

Answer

3 and 4

Question 4

The image-generation models marketed under the 'Nano Banana' name are developed by which one of the following Alphabet units?

Accepted Answer

Google DeepMind

Answer

Google Research, the company's core computer-science division

Answer

X, Alphabet's semi-secret moonshot research facility

Answer

Jigsaw, Alphabet's unit working on threats to open societies

Question 5

With reference to the fabrications produced using the Google Earth AI feature in 2026 and reported by open-source investigators, consider the following statements:
1. Scenes of refugees massed near the US–Mexico border were fabricated.
2. A fictitious nuclear facility in Iran was generated.
3. Because the generated images carried no watermark of any kind, they were entirely indistinguishable from authentic satellite imagery.
4. A hospital with a bomb crater in Gaza was among the fabricated scenes.
Which of the statements given above is/are NOT correct?

Accepted Answer

3 only

Answer

1 and 2

Answer

2 only

Answer

3 and 4

Question 6

In the context of the Google Earth episode, SynthID is best described as which one of the following?

Accepted Answer

An imperceptible watermark embedded into the pixels of AI-generated media, which persists through compression and resizing

Answer

A cryptographic signature written into the image file's metadata header, which is preserved through any subsequent re-encoding

Answer

A visible badge overlaid on a corner of every generated frame, which cannot be stripped away by cropping the picture

Answer

A public registry of image hashes maintained by model developers, which lets anyone match a suspect file to its generator

Question 7

With reference to the reliance of open-source investigation on satellite imagery, consider the following statements:
1. Satellite imagery is treated as a factual record in journalism and environmental monitoring.
2. Bellingcat used Google Earth-based verification to expose faked satellite imagery released by the Ukrainian government regarding the downing of Malaysia Airlines MH17.
3. Insurance assessments and legal disputes are among the fields that routinely rely on satellite imagery.
4. Humanitarian investigations and open-source intelligence work draw on satellite imagery as evidence.
Which of the statements given above is/are NOT correct?

Accepted Answer

2 only

Answer

1 and 4

Answer

2 and 3

Answer

4 only

Question 8

Consider the following as means by which a fabricated satellite image of a real place could still be caught by analysts, as noted in reporting on the Google Earth episode:
1. Comparing the scene against authentic imagery of the same location for inconsistencies such as misshapen buildings and distorted roads.
2. Validating fixed points of reference in the surroundings of a claimed disaster site.
3. Detecting the embedded SynthID signal, which persists through compression and resizing of the file.
4. Relying on the fact that cropping an image permanently destroys every embedded provenance signal within it.
Which of the above is/are correctly identified?

Accepted Answer

1, 2 and 3

Answer

1 and 4

Answer

2, 3 and 4

Answer

1, 3 and 4

Question 9

With reference to India's amended framework on synthetically generated information, consider the following statements:
1. It amends the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021.
2. It was notified on 10 February 2026 and came into force on 20 February 2026.
3. Significant social media intermediaries must deploy technical measures to verify a user's declaration about synthetically generated information before it is published, rather than rely only on self-declaration.
4. Synthetically generated information is defined to cover only audio-visual content and expressly excludes still images.
Which of the above is/are correctly identified?

Accepted Answer

1, 2 and 3

Answer

1 and 2 only

Answer

2, 3 and 4

Answer

1, 3 and 4

Question 10

Under the strengthened intermediary framework announced by the Government of India in August 2026, an intermediary that receives a notice from an appropriate government agency or a court must remove the unlawful content within how many hours?

Accepted Answer

3 hours

Answer

24 hours

Answer

36 hours

Answer

72 hours

Question 11

Consider the following statements about the platform now known as Google Earth and its predecessor:
1. It was first released in June 2001 under the name EarthViewer 3D.
2. Google relaunched the product under the Google Earth name in June 2005.
3. Google acquired the startup that built the original product, Keyhole, only after the product's relaunch under the Google Earth name.
Which of the statements given above is/are correct?

Accepted Answer

1 and 2 only

Answer

1 only

Answer

2 and 3 only

Answer

1, 2 and 3

Question 12

In January 2026, which one of the following countries was the first to block access to Elon Musk's Grok over AI-generated deepfake imagery?

Accepted Answer

Indonesia

Answer

Malaysia

Answer

Brazil

Answer

France

UPSC Prelims Practice Questions — How did Google’s AI satellite images raise safety concerns?

Q1. In its public statement withdrawing the AI image-generation capability from Google Earth in 2026, Google attributed the decision primarily to which one of the following?

Q2. Which one of the following user groups did Google itself name as the leading legitimate beneficiary of the Google Earth AI image feature even while announcing its withdrawal?

Q4. The image-generation models marketed under the 'Nano Banana' name are developed by which one of the following Alphabet units?

Q6. In the context of the Google Earth episode, SynthID is best described as which one of the following?

Q10. Under the strengthened intermediary framework announced by the Government of India in August 2026, an intermediary that receives a notice from an appropriate government agency or a court must remove the unlawful content within how many hours?

Q12. In January 2026, which one of the following countries was the first to block access to Elon Musk's Grok over AI-generated deepfake imagery?