Anonymous and Confidential Sources Policy

Principle

DesiAI generally prefers named and on-the-record sources because identification allows readers to understand a source’s expertise, access, interests and accountability. However, important public-interest information may sometimes be impossible to obtain or safely publish with the source’s identity disclosed. Anonymity may be appropriate when a source faces a credible risk of retaliation, loss of employment, legal exposure, harassment, discrimination, professional consequences, physical danger or another serious adverse consequence.

Anonymity is an editorial safeguard, not a matter of convenience. It is not granted simply because a source does not want to be questioned, wants to make an unsupported accusation without accountability or seeks to damage another person anonymously.

Scope and Terminology

This policy covers anonymous and confidential sources, whistleblowers, background briefings, non-attributable information, off-the-record communications, protected contributor identities, embargoed information, leaked material, source documents and personal information that could reasonably identify a source.

Terminology can differ between jurisdictions, industries and professional settings, so source arrangements should be clearly agreed rather than assumed. On the record means information may be published and attributed to the identified source. On background means information may be used under an agreed description without naming the source. Not for attribution means information may be published without identifying the source under the agreed attribution. Confidential means the source’s identity or supplied information is protected according to agreed conditions. Off the record means information is not intended for publication or attribution unless a later agreement permits its use or the information is independently obtained. Embargoed information may be prepared in advance but should not be published before the agreed release time.

Decision to Grant Anonymity

Anonymity may be considered when the information has meaningful public value, the source has direct or otherwise reliable knowledge, identifying the source creates a credible risk, the information cannot reasonably be obtained on the record, the source’s identity can be verified internally, the source’s motivation can be assessed, corroboration is available or realistically achievable, the information can be protected proportionately and the responsible editor approves the arrangement.

Editors should consider whether a less restrictive option could achieve the same purpose. Alternatives may include delaying identification, using a narrower description of the source, protecting only certain information, publishing documents without identifying the source, independently corroborating the information or anonymizing the subject rather than the source.

When Anonymity Is Refused

Anonymity will normally be declined when a source wants to make a personal attack, offers only opinion or speculation, cannot explain the basis of the information, is concealing a commercial client, seeks political advantage without supporting evidence or simply wants to avoid responsibility for an ordinary public statement.

It will also generally be refused where a source demands publication of unsupported claims, attempts to control the editorial outcome, requests payment for unverified information, materially misrepresents their identity, provides evidence that cannot reasonably be examined or asks for anonymity after publicly making the same information under their own name.

Seniority, fame, institutional access or professional title does not by itself justify anonymous treatment.

Internal Identity Verification

The responsible editor should ordinarily know the real identity of an anonymous source and verify relevant details such as current or former role, access to the information, relationship with the subject, chronology, motivation, financial or political interests and previous reliability, as well as the risks that may require protection.

Where an author knows a source’s identity but the responsible editor does not, material reliance on that source should normally require an exceptional arrangement and appropriate approval.

Identity verification should be proportionate to the circumstances. DesiAI does not seek more personal documentation than reasonably necessary, and records capable of identifying a protected source should be separated from ordinary article files where appropriate.

Basis of Knowledge

DesiAI should understand how a protected source came to know the information. This may involve direct participation, personal observation, access to records, professional responsibility, technical access, communication with a decision-maker, employment, official position, personal experience or documented secondhand information.

A person who heard an allegation from someone else should not be presented as having direct knowledge. The article should describe the source’s basis of knowledge as accurately as possible without creating an identification risk.

Motivation and Conflicts

Sources may be motivated by public interest, conscience, safety concerns, professional disagreement, political interests, financial interests, personal grievances, employment disputes, commercial competition, legal strategy, retaliation, reputation, ideology or a combination of factors.

Motivation does not automatically establish that information is either true or false. Editors should understand the source’s interests and test the information accordingly rather than assuming either complete good faith or dishonesty. Where relevant, a material conflict may be explained to readers without revealing the source’s identity.

Corroboration

Information provided anonymously should normally be tested against independent evidence, including another genuinely independent source, documents, data, recordings, public records, direct observation, technical testing, contemporaneous communications, physical evidence or an admission or response from the subject.

Two anonymous sources are not independent if one received the information from the other. Likewise, several employees repeating the same workplace rumor do not necessarily constitute multiple independent confirmations.

The published article should not imply a stronger level of corroboration than actually exists.

A single anonymous source may support publication only in exceptional circumstances where the source has direct knowledge, the information concerns a matter of substantial public importance, the source has demonstrated credibility, the risk of identification is serious, corroboration is limited for defensible reasons, surrounding records or circumstances provide meaningful support, the source’s motive has been examined, the subject has received a meaningful opportunity to respond, uncertainty is accurately reflected and appropriate senior editorial approval has been obtained.

Description and Explanation to Readers

A protected source should be described as specifically as possible without creating an unreasonable identification risk. Descriptions may identify a current employee with direct knowledge, a former official involved in a decision, a contractor with access to a relevant system, a person present at a meeting or an individual directly affected by a policy.

Descriptions should not make a source appear more senior, authoritative or knowledgeable than the person actually is.

When anonymity supports a significant claim, the article should explain the reason where practical. This may include risk of dismissal, contractual retaliation, legal exposure, physical danger, professional consequences or harassment.

The explanation should not combine seemingly harmless details in a way that allows readers or the subject to identify the source through deduction.

Source Agreements

Source terms should be agreed before publication wherever reasonably possible. The agreement should establish permitted attribution, permitted use, confidentiality expectations, treatment of documents and recordings, independent verification, future contact, timing, embargoes, security responsibilities and any meaningful limitations on the arrangement.

DesiAI does not promise absolute technical secrecy, immunity from lawful legal processes, control over the final article, deletion of accurate records on demand, payment for a favorable account, concealment of material deception or a level of protection that cannot realistically be provided.

An unsolicited message is not automatically confidential simply because it contains sensitive information or the sender describes it as confidential. DesiAI will nevertheless handle apparently sensitive information carefully while determining what protection, if any, can reasonably be offered. Sources should avoid sending highly sensitive records until an appropriate communication method has been established.

Payment

DesiAI does not ordinarily pay sources for information because payment can create incentives to exaggerate, fabricate or selectively provide evidence.

Reasonable reimbursement of necessary expenses may be considered in exceptional circumstances where it does not purchase the substance of the source’s account, receives prior approval and is disclosed when the arrangement is material to the published work.

Legal and Technical Limits

Source confidentiality can be affected by legal orders, compromised devices, service-provider access, metadata, account recovery systems, human error, malware, cross-border laws, unauthorized access, physical seizure, third-party disclosure or information that is already known to others.

DesiAI takes proportionate steps to protect confidential sources but does not represent confidentiality as technically or legally absolute. Sources facing serious employment, legal or security risks should consider obtaining independent professional advice. DesiAI is not the source’s lawyer, employer, personal security provider or legal representative.

Secure Communication and Document Handling

Communication methods should reflect the level of risk and may include separate addresses, encrypted messaging, encrypted email where available, secure file-transfer systems, avoidance of workplace devices, strong authentication, removal of identifying metadata, controlled access and separation of source identity from reporting files.

No communication method is completely secure, and sources should not be required to use technical measures they cannot safely understand or operate.

Source documents should be stored with access restricted to people who genuinely need them. Files may be examined for author metadata, revision history, hidden comments, file paths, usernames, printer identifiers, device information, location data, embedded objects and other details that could reveal the source.

Redactions should be checked to ensure that hidden text, metadata or recoverable information does not unintentionally expose protected material. Original evidence should be preserved separately from publication copies where necessary.

Internal Access and Draft Hygiene

Source identities should be shared only with people whose involvement is necessary for verification, editing, legal review, security, publication approval or source protection.

Commercial staff, advertisers and unrelated contributors should not have access to confidential source information. Protected names should not unnecessarily appear in public content-management fields, file names, titles, comments, revision histories, shared commercial folders, social-media drafts, email subject lines or routine analytics systems.

Confidential Sources and Artificial Intelligence

Confidential source material should not be entered into a third-party artificial intelligence system without appropriate authorization and a risk assessment covering provider retention, training use, access controls, international transfers, account security, contractual terms and potential exposure.

AI systems should not be used to identify, infer or reconstruct the identity of a protected source from confidential material or combinations of available information.

Source Review and Editorial Control

An anonymous source does not receive editorial approval over the final article. DesiAI may verify quotations, technical explanations, chronology, source descriptions, safety-sensitive details and information that could unintentionally identify the source.

A source may raise factual or safety concerns, but cannot require the publication to remove relevant contrary evidence or dictate the article’s conclusion.

Withdrawal and Source Deception

A source may later ask for information to be withdrawn. DesiAI considers the original agreement, evidence already obtained, independent corroboration, publication status, safety concerns, fairness, public interest, legal obligations and whether the information was obtained deceptively.

Withdrawal does not automatically require removal of independently verified information. However, a credible new risk to a source’s safety may justify delaying, modifying or cancelling publication before release.

If a source materially deceives DesiAI, the publication assesses what information was false, whether evidence was fabricated, whether the reliability of the article has been affected, whether other sources were compromised and whether a correction or retraction is necessary.

Source protection is not withdrawn casually or as punishment. Any decision concerning a source’s identity remains exceptional and should take account of safety, ethics, law, public interest and the original commitment made to the source.

Complaints and Whistleblowers

A person challenging an article based on anonymous sources is not entitled to the source’s identity simply because the reporting is disputed. DesiAI assesses challenges against the underlying evidence while protecting confidential information.

Where appropriate, the publication may explain the number and general category of sources, their basis of knowledge, corroboration obtained, reason for anonymity, documents reviewed and material limitations without revealing identifying information.

Potential whistleblowers may contact editorial@deshiai.in. They should avoid sending highly sensitive records until an appropriate communication method has been agreed.

Concerns that DesiAI improperly used anonymous sourcing or failed to honor a confidentiality commitment may be sent to complaints@deshiai.in.