Computer Ethics

Software is sold/distributed under different licensing categories. Cambridge expects you to keep these straight.

Free software (capital F or 'libre'). Users have the FOUR FREEDOMS:

1. Freedom to RUN the software for any purpose.
2. Freedom to STUDY how it works (source code is available).
3. Freedom to MODIFY it.
4. Freedom to REDISTRIBUTE (modified or unmodified).

'Free' here means FREEDOM, not zero cost. Free software MAY cost money. Examples: GNU/Linux, GIMP, LibreOffice.

Freeware. Software available at NO COST but with the source code KEPT CLOSED. Users can't modify or redistribute. 'Free' as in PRICE only. Examples: Adobe Acrobat Reader, Skype (older versions).

Shareware. Software you can use FREE for a TRIAL PERIOD or with LIMITED features. To unlock the full version (or after the trial), the user must PAY. Examples: WinRAR, classic Windows shareware.

Open-source software. Source code is publicly available; users can study, modify and redistribute (subject to a licence such as GPL, MIT, Apache). Often (but not always) also free of cost. Examples: Linux, Firefox, Blender, VSCode.

Proprietary software. Owned by an individual or company. Source code is CLOSED. Users licence it under restrictive terms — typically can't modify, can't redistribute, can't use beyond a specified user count. Examples: Microsoft Office, Adobe Photoshop, AutoCAD.

Cambridge tip. The single most-tested confusion: FREE vs FREEWARE. Free = FREEDOM (modify + redistribute). Freeware = no COST only. Don't merge them.

Copyright is a legal right that gives the creator of original work (including software, music, art, writing) EXCLUSIVE control over its reproduction, distribution and adaptation for a set period (typically 70 years after the creator's death in the UK).

What infringement looks like.

• Copying software without a licence.
• Sharing licensed software with others.
• Reverse-engineering for unauthorised purposes.
• Selling counterfeit software.

How developers protect copyright.

• Product keys / licence keys. Software requires a unique key tied to a purchase. Without it, the software won't run or runs in a limited mode.
• Online activation. The software contacts the developer's server to verify the licence is genuine and not in use elsewhere.
• Digital Rights Management (DRM). Built-in checks that prevent unauthorised copying or running. Common in commercial games and media.
• End-User Licence Agreements (EULAs). The contract users agree to when installing — sets out what they can and can't do.
• Code obfuscation. Making the compiled code hard to reverse-engineer.
• Watermarking. Hidden identifying information in software output.

Why this matters. Without copyright protection, developers wouldn't be paid for their work, and the software industry as we know it couldn't sustain itself. Counter-argument: overly aggressive DRM can hurt legitimate users (always-online requirements that fail when servers shut down, restrictions on resale).

Cambridge tip. Mark scheme expects students to know BOTH copyright (the legal right) AND the technical measures used to enforce it.

Personal data is any information that can identify a living person — name, address, email, phone number, ID numbers, IP addresses, biometric data, location.

Data-protection laws (e.g. GDPR in the EU/UK, similar laws in other jurisdictions) typically require:

• Lawful basis. Personal data can only be collected and used for SPECIFIC, lawful purposes.
• Transparency. Users must know what data is collected and why.
• Minimisation. Only collect what's necessary.
• Accuracy. Keep data accurate and up to date.
• Storage limit. Don't keep data longer than needed.
• Security. Protect data from breaches (encryption, access control).
• Subject rights. Users can ask to see, correct, or delete their data.

Why this matters. Personal data breaches expose individuals to identity theft, harassment, discrimination, financial fraud. Companies that breach data-protection laws face large fines (GDPR fines can be 4% of global revenue).

Cambridge tip. Mark schemes test students on the principles (collection, storage, use) and the user's RIGHTS (access, correction, deletion). Both halves matter.

Computing isn't free of physical impact, even though it can FEEL intangible.

Energy use. Data centres alone consume around 1-2% of global electricity (and rising fast with AI). Cryptocurrency mining and large-model AI training are particularly intensive. Always-on home devices (smart speakers, routers, modems) add up across billions of households.

E-waste. Short device lifespans (smartphones replaced every 2-3 years, computers every 4-5) generate around 50 million tonnes of electronic waste per year globally — and rising. Toxic heavy metals (lead, mercury, cadmium) leach from landfilled e-waste into soil and groundwater.

Resource extraction. Manufacturing chips and devices needs rare-earth metals (lithium, cobalt, neodymium, tantalum). Mining damages ecosystems, uses huge amounts of water, and often happens in countries with weak labour protections.

Carbon emissions. Tied to all the above — manufacturing, electricity, shipping, end-of-life processing.

Mitigations.

• Efficient hardware — newer chips are much more efficient per watt.
• Renewable-powered data centres — Google, Microsoft, AWS all targeting 100% renewable.
• Recycling programmes — recover rare-earth metals from old devices.
• Longer software support — let devices stay useful for longer.
• Fairphone-style modular design — easier to repair instead of replace.

Cambridge tip. Top-band candidates can identify SPECIFIC impacts AND name a mitigation for each. Treating environmental impact as a side issue is a common mark loss in 2026-2028 papers (the syllabus puts more weight on it).

Computing's recent acceleration in AI and automation has surfaced new ethical questions.

Bias. AI inherits biases in its training data. Famous failures: hiring tools that downgraded female candidates because historical hires were mostly male; facial-recognition systems that misidentified darker-skinned faces at higher rates; predictive-policing tools that over-targeted minority neighbourhoods.

Accountability. When an AI causes harm — a wrongful arrest, a denied loan, a misdiagnosis — who is responsible? The data provider, the model trainer, the deployer, or the user? Legal systems are still catching up.

Job displacement. Beyond manual labour, AI threatens cognitive jobs — translation, paralegal research, customer service, parts of programming, healthcare imaging. Society needs to plan for transitions.

Privacy and surveillance. AI makes large-scale surveillance cheap. Combined with facial recognition and location tracking, it can erode the privacy assumptions democracies have relied on.

Autonomous weapons. Lethal autonomous weapons systems can decide to kill without human intervention. Subject of intense ethical and treaty debates.

Misuse. Deepfakes for fraud, harassment or political manipulation. Personalised disinformation at scale. Phishing emails written by AI are more convincing than ever.

Cambridge tip. Mark schemes reward NUANCED views — acknowledging that AI offers real benefits AND raises real risks, and that responsible deployment matters.